⚠ This page contains old, outdated, obsolete, … historic or WIP content! No warranties e.g. for correctness!
News bottom
Note: all these entries are from tg@ up to November 3rd, when the new MirOS website was brought up.
29.09.2003
Release MirOS BSD #7r80, tag the tree. Build a full
release snapshot including X-Window and a ISO9660 file, upload it.
[a href="historic/ann-7.htm" | Read the
MirOS #7r RELEASE Announcement] for further information on this
release. There are known bugs: the Microsoft Windows emulators don't
work correctly, and several ports do not build at all, or only using
the MKC_EGCC=yes option to make(1). Several text
formatting utilities (troff, soelim, refer, ...) are broken or not
existent at all. We are investigating these for the next release, but
lack man-power. Volunteers are welcome to send unified diffs or gzipped
tarballs of patches that fix certain aspects of the system.
01.10.2003
Please read the release announcement again
after this weekend. MirOS #7bis will be released until then;
it is a critical bug-fix release only.
KDE 3.1.3 is playable (kmahjongg, to be exact); so both common desktop
environments of these days (not CDE though :( are supported.
03.10.2003
Sync with OpenBSD and NetBSD. Release Engineering.
04.10.2003
After fixing the whole stuff, MirOS BSD #7bis
was tagged, ports and source tarballs built on my own local machine and
we had a party in the Netzladen.
The base system including X-Window, as well as the predefined packages
(jupp and egcs) were built; everything checksummed and the ISO generated.
The webpages have been updated a lot. – Expect the files to be uploaded
and the anoncvs/cvsweb mirrors to sync until Sunday Evening, European
time zone. Have a lot of fun installing MirOS BSD #7.
05.10.2003
Errata: /etc/rc fails to force-mount
UFS ffs filesystems when /usr is on a separate filesystem. An
error message is output, and in case of unclean filesystems and disabled
fsck(8), the system fails to start up.
A fix for this, re-implementing
the sort(1) program in Korn shell, has been committed to the CVS repository
and MFC'd back to the MIRBSD_7 tag. The etc34.tgz tarball on the
download site has been updated. – Touched files: /etc/rc.
06.10.2003
Errata: Same file, same bug, different executable.
Nuke use of /usr/bin/sed, replace with /bin/ed in the
file /etc/rc. Update tarball again.
07.10.2003
Errata: Same file, same bug, same fubar.
08.10.2003
Finishing the MFC spree yours sincerely proudly presents
a new set of source and ports tarballs – this time with corrected
permissions, CVSROOT, Tag and rc files. Due to hardware problems, the update
of the anoncvs server will take a bit longer this time. — Also end
argumenting with the ports people from OpenBSD, not removing their names but
circumventing an argument until the solution is done on November 1th. Benny
Siegert builds new packages for GNU GNOME et al.
09.10.2003
Well, the whole shit about
/etc/rc has one good part: I've learned a huge lot more about
ksh(1), today's features: co-processes.
14.10.2003
Still slacking – well, relaxing. I need this from time to
time. In the meanwhile, Waldemar Brodkorb of www.OpenBSD.de fame helps
Benny to mop up the whole ports tree, from A to Z.
20.10.2003
Catch-Up listing: recover, try to get vacation (and fail).
Do a bit of "webdesign" (though I don't like the WWW at all) and prepare for
a new MirOS homepage (I've already finished a design study, and it's all
XHTML 1.1 superstrict, lynx optimized, displays well in Bloatzilla and even
Konqueror). Benny goes whack on the MirPorts framework again, with help from
Waldemar Brodkorb and Angelo Laub, as well as some other people reporting
breakage. It's said that boehm-gc is working now. (Mozilla and Firebird are
still BROKEN on MirPorts, not only because of gcc hard-coded, but also
because it doesn't work (either natively, nor the OpenBSD package, nor in the
Linuxulator). Fix (really: work around) a bug in the pckbc code (well,
better: the broken BIOS and hardware (8042 controller) of some broken Laptop
vendors) which caused the keyboard to eventually "freeze" (it fails to attach
correctly, yet shows up in dmesg) – it sends a false response to the RESET
command, breaking the protocol (we can cope for that – but warn the user,
since if anything breaks, it's not our fault; they should
move the laptop to /dev/bin anyways). Make up some ppp.linkup and
linkdown sample files which fit MirADSL and MirISDN much better. Communicate
with Hellmuth Michaelis from ISDN4BSD, who's _really_ lucky that, finally,
someone's supporting it for an OpenBSD-alike platform. – Oh, and I'm finally
mentioned in the famous Unix family chart,
although some of the lines are incorrect or even missing (I forked from
OpenBSD and NetBSD right away, and took fairly some
(legal) MicroBSD code) yet.
21.10.2003
Collect mailing lists: ports -> miros-discuss;
*-changes -> miros-changes. Welcome new developer wbx (Waldemar Brodkorb)
who volunteers to maintain the sparc32 port of MirOS BSD. Cleanup in CVSROOT.
Nuke ctm port, it's in base since 3 releases. Import boehm-gc from
OpenBSD-current ports tree. Port over GNU GRUB (pxegrub, nbgrub, stage{1,2})
with netbooting support for MirBSD kernels; remove a.out pxegrub/bsd.rd from
tree.
24.10.2003
MFC the workaround for broken pckbd. Split the tree (branch
tag MIRBSD_7_DEV (leading to MirOS BSD #7ter) starts at MIRBSD_7 (released as
MirOS BSD #7bis). Sync NetBSD acronyms{,.comp}. Nuke OpenBSD ISC-style
license template, re-word application notice of MirOS MIT-style licence.
Import NetBSD lib/csu into vendor branch and merge changes for common_elf as
well as i386 and sparc. Sync bc and dc with OpenBSD-current on both branches.
Bring back the sparc(32)-related stuff from OpenBSD. Fix netris and aspell
ports. Move ATM and PLIP defines out of machine-independent part of GENERIC.
Nuke bsd.nls.mk include. Adjust X for sparc support (probably).
Change OpenBSD to MirBSD in installation instructions.
26.10.2003
Build up my SPARCstation 20 at home and have it compile
a GENERIC kernel. Bring the port to sparc32 that much back to work that it
actually seems to do something useful (this includes synching three files
to OpenBSD-current, coding and inventing some stuff myself, and the usual
bring back, try, fix cycle). Fix DNS and make up a third nameserver for my
domain space, at dynamic IPv4 like the others :(
27.10.2003
Large-scale cleanup of the main tree as well as the stable
development tree. Make i386-HEAD and sparc-DEV build. wbx@ whacks USE_CXX on
a huge part of MirPorts. Harden OpenSSH-MirBSD (no root login by default) and
fix documentation. Add RC5 and MDC-2 algorithms to libcrypto, since they are
only patented in the United States of America. Fix usr.bin/{bc,dc} mergo.
Discover and fix most of sparc breakage. Import gcc-3.2.3 partially and nuke
non-free documentation and some not used stuff.
02.11.2003
Having fulfilled my duty to remove all peoples' names off
MirPorts, play with other stuff (random, cdboot) – contact the ChangeLog for
details, I'm too exhausted to play www tonight.
03.11.2003 by tg@
Switching to the "new-style" website, making old
content still accessible though. Switch most distfiles to bsdadvocacy.
distfiles now belong into HERC:/cvs/distfiles/ (open for ports ci).
07.11.2003 by tg@
Allow mknod calls within a chroot (enables us to
build releases within a chroot jail). Remove contributed OpenBSD songs,
since They obviously don't want "us" to be an OpenBSD patchkit any more.
Optimize /etc/profile: show a progress bar, unbloat a little; use less(1)
as default pager. Actually build the MIRBSD_7_DEV tree through on a sparc.
Fix default warnings of /etc/security by installing files (such as
inetd.conf) and directories (such as /var/mails) with the ownership and modes
that the security script expects. Update from Apache HTTPD 1.3.28 and the
OpenBSD libssl/libcrypto (based upon a bastardized OpenSSL 0.9.7b) to Apache
HTTPD 1.3.29 and OpenSSL 0.9.7c, retaining OpenBSD and MirOS local changes.
Revert evilwm to the old resize/move behaviour to allow for one-pixel moves.
Finally fix the automatically generated subdirectory index for Apache with
the MirOS variant of "404 Is A Directory". Merge OpenSSL CA Certificates
into what is installed by default. Re-enable RC5 and MDC-2 (only patented
in the USA). Help with mdoc error messages by printing out a kind of real
estimate where in the file the error lies (XXX need a regression test!).
Rewrite rand() and srand() functions in the kernel from scratch, taken from
my Netfisch cipher implementation (rand, i386 assembly). srand() now doesn't
set the seed, but rather make the new seed a function of the old seed and
its parametre, so nobody ever knows the exact state of the PRNG seed of
/dev/prandom and rand(). The PRNG is seeded with arc4random() every 256
seconds, so it's a pure random generator now. ANSIfy, KNFify areas being
revisited, as well as addition of the MirBSD RCSId and removal of white
space at end of line/file, and the character 0x60. Change the SLIM kernel
to be pretty much what is used on HERC. Enable 'make show=FOO' for the
base system. Move bootblocks to a "better" directory structure. Have
deprecated gas pseudo-op '.file' die. Add cdboot which works like /boot
but reads its kernel directly from an ISO9660 filesystem, and must be
burnt as boot image in "0x00: no emulation" mode (XXX the cdboot program
does not correctly work at all YET). Move boot loader version number to
/sys/arch/i386/stand/Makefile.inc, so it can be bumped in one central place.
Kludge mkisofs from cdrtools-2.01a19 (newer than ports/sysutils/cdrtools)
into the tree (XXX it's REALLY a kludge, and bloated), nuking mkhybrid
which did not support "0x00: no emulation" mode (nor HDD emulation, for
that matter, at least not correctly). Add the "hacker emblem", the slider
from John Conway's 'Game of Life', to /etc/motd and nice-i-fy that file.
Rename /etc/boot.conf to /etc/boot.cfg (needed for cdboot to function
as intended). Simplify and speed up /usr/libexec/randshuffle. Let gencat(1)
die, it's an unneeded NLS relict. Kludge the new-style website to work with
the outdated php4 we got on mirbsd.bsdadvocacy.org, fix XHTML/1.1 validation.
Remove every and any GNU FDL licenced file out of the tree (XXX this breaks
the tree build, not fixed until today YET), because you already breach that
licence if a file is not world-readible or saved on an encrypted filesystem.
Add a command-line parametre to ftp(1) to restrain it from using EPSV on
IPv4 sockets (some ftp dæmons really don't like it and have us wait
for a connection timeout, as much as 30 minutes and more). XXX: we NEED a
key handler which sends a signal 1 to ftp (itself). Apply OpenBSD errata
which are needed to the HEAD branch. Enable nroff(1) output with -Tcol
(-Tascii, -Tlatin1) to be piped through col(1) yielding multi-column
output, and add nrcon(1) script which does that automatically. Switch
<bsd.doc.mk> to use nrcon, not nroff. Update lang/lua to 5.0 and
devel/libtool to 1.5 (lua 5.0 and libtool 1.4.x are needed for ion-current),
and ion to 20030814-devel, asked for by Tim Kornau. Add (commented out)
hints to bktr(4) to the GENERIC kernel, in case it does not correctly
autodetect the TV card (from Robert Nagy). Clean up after wbx@'s new ports.
Rename 'gs' function (in /etc/profile) to 'ggs' to avoid clash with
ghostscript.
07.11.2003 by bsiegert@
Take over misc/ccze and fix build. Fix up
archivers/rar (BSD/OS emulation). Update net/mldonkey to 2.5-4, with bug
fixes, and polish/clean up port. Import gnome-themes-extra and gnome games.
Apply OpenBSD 3.4 Errata patches to MIRBSD_7_DEV branch (asked by tg@).
07.11.2003 by wbx@
USE_CXX, USE_X11 whack on MirPorts. Make numerous
ports build correctly, or at least fail gracefully. Fix sparc boot-ISO
generation. Update www/snownews and net/pptp. Import graphics/xawtv. Change
eMail contact address. Add gtk flavour to audio/mp3info.
08.11.2003 by tg@
Catch up website log with reality. (That would be
the three entries above.) Move GNU docs to src/gnu/doc and simplify; add
ancient man pages for as(1) and ld(1) while here. Build system through, fix
showstoppers and sync distrib set lists; bump patchlevel. Update Apache HTTPD
default configuration with a good shot of reality drugs. Speed up the MirOS
homepage's PHP processing. (At least according to www.php.net information.)
Move a lot of the old homepages' content into the "new CMS" of MirOS.
Prepare for the (unsupported by MirBSD) "new" OpenBSD pkgtools which might
get imported later. Re-run autoconf on the tree and fix breakage occured
in GNU as (once again, *sigh*). Continue working on the CD boot loader (which
is almost that ready that it can boot a kernel from CD; though neither /boot
nor cdboot.bin are usable at all at the moment (yes this means I broke them).
They'll be completely reshaped later.
09.11.2003 by bsiegert@
Add patches for MPlayer to use ~/.etc/mplayer.
Update BIND9 port, you can now install only the client tools.
09.11.2003 by tg@
After many prodding of wbx@ work on the MIRBSD_7_DEV
branch again: remove OpenBSD elf2aout and elf2ecoff, and move to NetBSD
elf2aout (with KNF and ANSIfication and whatnot), allowing for bsd.rd.net to
be made. Physically update my SPARCstation 20 to a 75 MHz CPU, closing the OpenBSD
problem report for me. Re-enable ramdisk for sparc, and, while here, add
back miniroot. Back to the head branch, remove man4/mvme88k and other unused
stuff like that. Strive to update X-Window and remove the OpenBSD dain bread
ness (to remove so-called "bad, copyrighted, licenced" pictures of that Tux
thingie, and our dæmon). Do optical and Validator fixes to the
homepage. Lose edonkey, the distfiles are gone and we ain't allowed to
mirror.
12.11.2003 by tg@
Have ports in BATCH=Yes mode be IGNOREd when they
depend on a USE_CXX or USE_X11'd port which is not yet installed, but on
IGNORE as well (recursive up to the end); but don't check dependencies on
myself (not recursively checked). Fix other stuff in MirPorts while there.
Update and take over graphics/win32codecs. Change definition of NULL. Enable
lots of more stuff in GENERIC and test until it compiles again; fix several
severe bugs which have been catched due to new NULL (including one in UVM!).
Add tun0 to MAKEDEV for ramdisk. Fix mk.conf REL_NO_KERNEL. Sync
sys/netccitt with NetBSD (XXX needs tweaking, but not as much as OpenBSD one
with dtom). Mail marc.info again wrt. archiving the MirOS
mailing lists.
16.11.2003 by tg@
Unbreak audio/lame. Synchronize master.passwd and
group with what MirPorts expects (MFCd). Update magicpoint. Fix ISDN manual
pages, and refer to MAKEDEV(8/i386). Switch boot/cdboot and cdrom to the new
format: /usr/mdec/boot is now both an ELF executable to be loaded
by the first stage boot loader (/usr/mdec/pbr_*), as well
as a flat executable to be loaded by the El Torito BIOS implementation (yes
this is possible :-) in "No Emulation" mode. It is capable of
loading a kernel from floppy, hard disc or the El Torito accessible CD-ROM
session (in 8.3 filename mode only), from UFS FFS filesystem as well as ISO
9660. The bsd.rd kernel has to be either stripped or gzipped in
order to be booted from CD-ROM (the new cd34.iso generation code takes care
of this) for some unknown reason, though. Add placeholders for recent and
future NetBSD, OpenBSD and MirOS releases in the manual page sheets. Update
acronymes database. Import OpenBSD ports tree into MirPorts. Work on the
documentation of MirOS (for the upcoming release of MirOS BSD #7ter). Fix
and enable X-Window in 7DEV. Make gas shut up. Add a knob which turns off
all (known to me; IANAL) patented algorithm for outside Europe (stock MirOS
is Europe-only and includes algorithms such as RC5™ patented e.g. in
the USA) to /etc/mk.conf. Add basic vmware 3 port (needs testing), and fix
modload. Update win32-codecs and mplayer. Add new RTL8029 clone. Make all
ports which USE_CXX and are built with EGCS depend on libestdc++ (package).
EGCS is i386 only. Go crazy on the website, and fix XHTML too.
17.11.2003 by tg@
Deprecate all calls to bzero(), bcmp() and
bcopy() as NetBSD has done. Whack more on the website; add info to
Éric Lévénez (here and on his Unix history page)
and resubmit MARC. Fix PBR after bringing it down to be incompatible to
prior-2.50 /boot, and shrinking by ~20h bytes. Have PBR and installboot
wear their version number.
19.11.2003 by tg@
Bring back old libtool. Verify that kernel and
boot tools still build and are usable (in bochs). Fix audio/faad and
x11/mplayer (which has faad enabled by default now); add no_faad flavour
to mplayer (for these guys on SPARC, where is no C++). Add incomplete
ethertap device framework (need to work on the .c file now). Inspired by
the CCC guys, fix up /etc/profile including the colour/bw and
prompt selection logic. Fix elf2aout and notice it doesn't do its job
cross-platform. Screw the i386 and sparc kernels in DEV. Remove RSAv1 and
DSA keys from the status of being enabled by default; fix and enhance the
afterboot(8) manual page while there. Prepare double-build-through of the
MirOS BSD #7ter release (borrow a VT for my sparc).
20.11.2003 by tg@
The usual problems with make b-r of 7ter
appear; go after them and try to fix. Meanwhile, add /kern and /proc to
the head branch and have the installation script amend them with meaning.
Ensure the kernel in its current state builds, and eliminate all "M "s
in my local tree, to facilitate backing up the whole stuff on CD again.
24.11.2003 by tg@
Mirror the fact that we have released MirOS #7 for
the third time on the website, after all uploading (and checksumming) has
finished. Renumber my IPv6 space (argh, why?).
29.11.2003 by tg@
Catch-Up entry again, I had no time this week.
Nuke bcopy(), bcmp() and bzero() from boot code and memcpy() ->
memmove() which I inlined — and take that inlining back, since
increasing the size broke cdboot again. Remove the b*() functions from
src/lib/ and make it compile with -Werror, as well as clean much of the
code for -Wall (but not everything yet!). Make tree build and release
cleanly. Remove creation and parsing of /etc/mygate and advise
users to use hostname.if(5) instead. Make gcc3 compile (but nothing more;
even the install process is more than broken) with target
i386-ecce-mirbsd3.4. Plan to change the output of uname(1) to
MirBSD instead of OpenBSD (idea from ekkoBSD).
Make gcc-3.2.3 (C,
without propolice) build and being used. Fix the string functions in
GNU libiberty.
30.11.2003 by tg@
Succeed building through the whole system using
gcc-3.2.3 without propolice (kernel, userland, X-Window). Make port of
GFDL licenced documentation. Move distfiles etc. around. Find out that
the boot loader is broken, but not in bochs (wtf?). anoncvssh
now handles anonRsync as well, using rsync-static flavour.
01.12.2003 by wbx@
Added verified exec from NetBSD 1.6.1 /
Stephanie for OpenBSD 3.4 (code by brian at ethernet.org, great thx)
to MIRBSD_7_DEV, one bug have to be fixed later.
Today I have began with an MirBSD port for alpha architecture.
01.12.2003 by bsiegert@
(Catch-up entry) Add db4 port and run
the regression suite which took almost two days to complete. FWIW, the
log is here. Work on
apr and apr-util (still unfinished). Re-install frankie alias
benz.mine.nu with MirBSD #7ter and give wbx@ an account.
Sparc packages are
now hosted there too. While here, install djbdns. Fix missing Fontmap files
for cups-pstoraster. Create a CD with packages and distfiles for templis.
01.12.2003 by tg@
Seems as I'm late today. It's still pretty obvious,
since I had the rope to shoot myself into the foot yesterday, and to cut me
into the thumb the day before yesterday (literally, this time). To be exact,
I dared to install /boot v2.53 (gcc-3 compiled) on my laptop. Well, real
hardware is not like bochs which is which I test with. Real hardware
sometimes returns errors when it shouldn't. But then, I fixed it eventually
(else I couldn't have written this), and having discovered /boot can soo
easily be called from DOS, I've written up a MS-DOS quickstrapper which
comes handy if you got a kernel on /dev/wd0a but don't know how to boot it.
If someone would now come up with a port of disklabel(8) for MS-DOS, you
could (using e.g. Norton Utilities 4.5/DOS as dd(1) replacement) easily
bootstrap a MirOS system on such boxen. Hm, and www.bsdadvocacy.org has
gotten downtime due to an update to OpenBSD 3.4, so I could switch on (and
fix) the "content/foo.index as it should have been from the beginning".
Next thing on my TODO is to too generate a CD-R for this templis guy Benny
wrote about, but I think I'll watch anime, go to bed, and do that tomorrow.
Coming from $ork, I'm pretty exhausted again – plus I've had a 25-hour
without break Netzladen session again (meal breaks excluded of course).
Hey cool, I think I'm gonna win the "who writes most into a wlog message"
pissing contest (I'm already equipped with the largest commit message ever
one). Oh, and mop up the CVS repository (one by me, one by wbx). Ah, and
set up a cronjob so that the CVS repo copy on mirbsd.bsdadvocacy.org is
automatically synched overnight (European time).
03.12.2003 by tg@
I've actually verified that one can build a CD
which is bootable on both i386 and sparc, using the MirOS #7ter release.
I'll have to research further in why my Laptop can't boot the no-emul
boot loader but the Win2k one. Now, I've synched the OpenBSD Ports Tree
and much stuff of base (bc, dc, perl, binutils, vnd) into MirOS/MirPorts;
you'll need a re-run of /dev/MAKEDEV after running a new kernel.
04.12.2003 by tg@
Finish merger of OpenBSD; build system through
and sync distrib set lists. Nuke a.out, iBCS2 and SysVR4 compatibility.
Split syslog.conf in MI and MD part. Apparently fix mod_perl
due to our IPv6 Apache HTTPD. Get annoyed by wbx@.
05.12.2003 by tg@
In real life, it was a wonderful day, until I
got pissed off by some lusers. I'll separate now cleanly between hacking
on MirBSD (which is my hobby) and helping lusers (which I get paid for).
If you want me to help, pay me. Isn't that simple?
Propolice works;
the switch to gcc-3 as system compiler was done. Start planning the next
stages. Get (probably useful?) feedback from this Templis guy. End life of
80486DX support in kernels and sync configs with each other.
09.12.2003 by tg@
For wbx@ in absentia, fix licence of vexecctl.
10.12.2003 by tg@
"Fix" (for some low values of fix) libtool (for
even lower values of software) good enough so base builds through for
C and C++ with gcc-3.2.3-propolice. The system compiler switch is deemed
done; sparc, fortran, objective C, probably Ada will follow after an
upgrade to 3.3.3 or so.
12.12.2003 by tg@
I'm still physically not well – my head aches from
time to time, and my stomach likes to fight with the head about who's best
in annoying me; my nose's either running or fully closed. Despite this, I
could finish and upload some -current snapshot for i386 (compiling doesn't
need much attention, just time) and fix wbx' Makefile for vexecctl in 7DEV
(one-liner...).
A word to my trainer at Deutsche Telekom if you
are reading this: all that went yesterday or overnight, and wasn't the cause
I'm ill today. Wish you have fun at the restaurant.
14.12.2003 by wbx@
Updated the openldap port, after some hours of
libtool debugging (set -x is a nice thing) it is working fine.
compiled all related packages successfully and tested most of the important
ldap related stuff with openldap 2.1.25. (gq,login_ldap,net-ldap,sylpheed,
pine-ldap, pure-ftpd,auth_ldap,...). SASL FLAVOR is broken, I need to
compile openldap and cyrus-sasl2 with the same libdb version (openldap use
libdb4, cyrus-sasl2 libndbm).
Updated silc-client port to 1.0, I will try the irssi-plugin next.
Security update of irssi-port from ports@openbsd.org
16.12.2003 by tg@
Fix screwup done to this log by wbx@. Various build
fixes and improvements. Fix gcc3 option -fno-no-stack-protector (sic!). Add
a regression test for propolice. Add new users and groups from MirPorts.
Call back from the Dead my old freenet-project port (which can't connect to
"the" freenet because our version is too old, and we can't update to
sunjdk-linux-1.4.1 because that doesn't do green threads any more – anyone
want to do a rogue freenet2? :). Make java build by the way. Add workarounds
to ksh extensions in /etc/profile because GNU bash sucks.
17.12.2003 by tg@
Let the perl port make fragment check the MirOS
patchlevel to determine if perl 5.8.2 is already there. Move old samples
from the time when MirBSD™ was my private project out of the tree,
into the contrib module, and make a port for that. Employ spelling fixes
and an update to the licence template and the COPYRIGHT file. Merge MirBSD
#7ter into the HEAD branch (not expected to work). Start working on case
studies and rapid prototypes to redesign the El Torito boot code with
something that works around buggy hardware. (Why am I still a bit ill?)
21.12.2003 by tg@
I'm the only one to write here ;-) hmm... anyways.
Enable libmilter by default and provide neat ways to enable LDAP and SASL
support for sendmail in /etc/mk.conf. Update HERC to #7ter with
the SLIM kernel configuration file from latest -current. Invent a makefile
include for MirPorts, <bsd.port.ver.mk>, to be used for fine-grained
versioning of ports, removing the need to branch MirPorts. Fix the boot
loader so it runs straight out of the box as /boot called from a v2.64 pbr
(older ones won't probably work), or as boot.com from MS-DOS (with not
much programs loaded), or as "no emulation" El Torito boot
image on a computer with a die-hard broken BIOS. Update the website, point
better to licencing information in-tree and on the site. Catch up missed
stuff when merging #7ter into HEAD. Finally fix the bootloader,
so no more complaints allowed for v2.65! Remove reference to
orphaned databases/db4 port noticed while doing make index.
25.12.2003 by tg@
Make and commit new ports index; remove netscape
for sparc which needs SunOS emul (still more to follow). Change uname
from OpenBSD to MirBSD and nuke -M option; fix manpage etc. Remove some
unpolite words about the OpenBSD people I forgot to before in this website.
Import OpenBSD and start to merge content over. Update and fix (some of
my strlcpy() stuff) libiberty; MD5 and getopt* are in libc. Enhance and
amend sixxs-heartbeatd samples, idea by bsiegert@. Merge in and MirOSify
more of the import (the whole sys/ stuff and libz).
We were going to
hold a speech at 20C3, but the people from CCC Berlin are (according to
insider information) lazy bums smoking shit, so they could tell us within
more than three weeks neither that nor that not we can hold that speech,
or even when. That's why we aren't interested in holding a speech any
more (well, at least I personally am not).
Set up a rogue vserver
for the MirOS website, in case the other one gets lost and we need to
tell people at the congress. While here, credit all the guys.
26.12.2003 by bsiegert@
Another catch-up entry, written in the train
to Berlin: Help tg@ with ports import from OpenBSD. Import security fixes
for rsync and screen. Add my db4 port, moving db3 to databases/db/v3. Roll
back those changes a few days later, remove db3 entirely and fix the two
ports that depended on it. While here, upgrade db4 to 4.2.50 and remove
compatibility kludges.
Now that subversion builds cleanly, import it into the tree, making a mess
during the import. Create an experimental subversion repository of MirPorts
at benz.mine.nu, accessible only via ssh, contact me if you want an account.
Fix dependencies of bitlbee. Request an IPv6 tunnel at
SixXS and configure a static IPv6
address for benz.mine.nu (not in the DNS though). Download CD image from tg@
and burn a few MirBSD and MirPorts CDs for 20C3. Go there.
26.12.2003 by tg@
Sleep longer than expected, then drive to
Berlin for the 20C3 (crappy website)
with wbx@ and IsCs from CCC Köln. Meet bsiegert@ there and many
other BSD{,s'} guys; have a nice evening with them and Club-Mate (just
the food wasn't at top level, neither is the network working, thus I've
made a HostAP (called mirbsd of course) for the internal communication
(even the APs are working so badly) of ours, eg. to view a local mirror
of this wonderful website :) Then start hacking on the source again,
together with bsiegert@, and talking to people. We're now dragged into
going to bed though...
27.12.2003 by tg@
1. half of the day spent fixing this MirBSD
uname issue, and the merge conflicts. Also, notice that anonCVS is
working again, as well as the 20C3 internet access (sort of; I'm still
providing a rogue access point – which I must now take down while
writing this – making the overall not-a-network situation worse...).
Joke with henning@ and wvdputte@, get interviewed by hubertf@netbsd
and josef@freebsd. Build a new total snapshot of base !X11. Quickfix
the uname problem for most ports, and check bind9 for benz.
27.12.2003 by bsiegert@
Fix bind9.
28.12.2003 by tg@ – well, sort of
Hum, this was a weird day.
Much happened, but I mostly remember having eaten well at the indian
again, but with benz, waldi and three freebsd people this time. Also,
I met tbf from symlink and some others (Immo FaUl Wehrenberg, and a
couple which I don't remember right now). Also I installed MirBSD to
people, but as for donations... they aren't going well, and these
OpenBSE people are still annoying :P Well, and some minor fixes.
29.12.2003 by bsiegert@
Meet lots of people from IRC, celebrate
winning the shortest-wlog-message contest with my last entry :P. Add a
port for the cvs version of dillo which fixes an annoying bug. Fix
hfsplus (well, sort of) although it doesn't run on any of our supported
arches. I am hacking on my iBook here with OpenBSD and MirPorts on
it.
30.12.2003 by tg@
We've driven home overnight, since congress
ended early (some foreigners are having a new year's eve party now,
so the location had to be left and cleaned up at 09:00 UTC today).
All in all I must say it's a super event, and we will take place in
it next year too. We've met a lot of people we already knew IRC, and
a bunch of people we didn't know, IRL – and some of them will come
into our IRC now as well. We didn't sell much MirBSD, but donated some
CDs for the Hacker Jeopardy to spread it (though some people just were
happy about the jewel cases) and chatted with visitors about it. Also,
we stopped the OpenBSD people that attended from murdering us ;-)
Today, I'm opening a new CVS branch tag MIRBSD_7 mimicking the
OpenBSD -stable branch (OPENBSD_3_4 as opposed to OPENBSD_3_4_BASE which
is our MIRBSD_7ter tag). Benny has volunteered to maintain and persuaded
me to ok the MirOS #7-stable branch, and we will continue to do so for
later releases. Also, in contrast to OpenBSD, we will provide binary
upgrades as site34.tgz unless there are changes which make it easier to
release a new full set of distfiles.
Attempt to port makefs(8)
from NetBSD (failed: the generated files, even after disklabel, lead
fsck_ffs(8) to floating point errors and force it and mc after mounting
to dump core. Try to fix XFree for uname change (needs revisiting).
Later: play X-Window and even get rid of the manpage generation errors
(by using a kludge though – anyone else uses not GNU groff?).
30.12.2003 by bsiegert@
Finally fix those stupid punctuation errors
in the install files. Fix the bind9 installation script (this time for real)
and do further testing. Enable building of threads flavor.
20C3 is over, and it was a great success. We even sold a few MirBSD CDs and
helped some people install it. I met lots of nice people and had lots of fun
hacking MirPorts in the hack-center. Now I am sitting in the train back home
:(.
02.01.2004 by tg@
Renew OpenSSH public key. Fix X-Window compilation
(except that fonts seemed to crash my machine, but that might have been
heat; now that I could use sensorsd(8) I probably should). Make ocaml and
mldonkey compile under #7. Port the "LIVE.COM streaming libraries"
for mplayer to be able to watch rtsp streams (just that I can't test, being
behind a NAT). Update mplayer from OpenBSD to 1.0pre3 while here, and make
it compilable without C++ and X11. Regenerate
ports/infrastructure/db/config.* GNU tools automatically and invent an ed(1)
script to copy and mask the RCS Id. Write a simple database (integer, string)
on top of libc-db-1.85 and the recno interface; adjust MirPorts pkgtools to
match recent OpenBSD Ports changes with that. Extend our copyright to the
year 2004. Start working on the mirDePinguinator (remote MirOS install tool).
Import OpenBSD base system and merge conflicts. Remove Kerberos and related
pieces totally from OpenSSH. Start preparing support for disklabel partition
in an extended BIOS partition, requested by Jedi/Sector One. Get request by
Diana Eichert to port NetBSD pxeboot over. Split a c++ subpackage from
gfdl-doc port.
03.01.2004 by tg@
Merge OpenBSD fully over. Build the base system,
sync lists and ramdisk sizes. Rename the ramdisk M and CD kernels to FLOPPY
and CDROM for i386. Fix INSTALL.i386 a little bit. Fix ports infrastructure
for #7 on sparc. Enable moving disklabel partition into an extended BIOS
partition. Rotate ChangeLog, write new wlog entry (hrhr).
06.01.2004 by wbx@
Tested the new ipv6 capable installer which
actually works fine. (configuration could be made permanent after reboot,
by adding rtsol to hostname.if and modifying sysctl) Doing some kind of
simple work for tg@, search for bcopy/bzero and change it to memmove/memset.
Added two new ports: samba3 and sendfile.
I've worked a lot on porting smbfs from NetBSD to test my new samba3 port,
but this seems to be not a simple task. Too many differences between NetBSD
and MirBSD.
07.01.2004 by bsiegert@
Attempt to do a full bulk package build and
fail miserably. The bulk building parts of the ports framework seem to be
broken, e.g. spitting out messages like Fatal: no flavors for this
port etc. What's more, egcc on #7ter often crashes. And somehow, the
linker is unable to find shared libraries in the ports' working directories
if building in chroot. This breaks building (among many others) glib2,
because it calls a program linked against itself while building.
Okay, I have 1289 fluffy unzels (to be uploaded IDC) in in my
packages/i386/ftp directory. A full build might be due next week
when my new hard drive will arrive :). A propos hard drive: There is going
to be some downtime for benz.mine.nu
while installing the new drive.
Fix some of the build errors encountered during the bulk build. Try to
incorporate Fefe's IPv6 patches for
ucspi-tcp, only to find out that they don't work for us.
08.01.2004 by tg@
Fellow miros-changes@ readers might have noticed
I slacked a bit the last week. In fact, I think I need holidays, since the
only two weeks of non-dayjob-work I had weren't very recovering (but still
fun, nevertheless). Despite that, I have done quite some work, although
there are no major changes at the moment (we're working on improving the
MirOS infrastructure). Anyways, this is what I've done:
Add pgp key
fingerprints and ssh public keys of the developers to the mirex contrib
repo (preliminary, not official yet). Find out that my diffs to disable
I18N/NLS/LOCALE for X-Window actually work, and – on top of that – make
the X-Window build subsystem recognise our $CC, $CXX and $COPTS while
building. Add ping6, rtsol and sysctl to the large ramdisk – as wbx@
wrote above, it makes install in an IPv6 network pure joy. Remove gcc
and libstdc++-v3 testsuites from the tree and remove them, as
well as the imported OpenBSD gcc3, hard from the repository.
Take back allowance for installboot(8) to write itself into the MBR,
because the code is incomplete (it wouldn't be hard to fix and test it,
but I have set priorities differently at the moment). Fix all the MDOC
warnings encountered during a build. Rewrite the tools in
src/X-Window/distrib/sets/ because they had no licence, and
move them to src/distrib/sets/ (effectively nuking and replacing
the OpenBSD base install ones, too) – they work even better now, path
independently and save us 4.2 KiB in the base system sets (and a few in
X sets). Remove duplicate introduced by OpenBSD merger in sample
openssl.cnf installed in /etc/ssl. Fix acroread port; we are
not OpenBSD. Throw in a warning into /etc/mk.conf stating that
doing a make release is not supported when changing flags.
TODO:
Employ real fixes for the installboot code, wbx' problem with the automatic
generation of the access and aliases database during a release, and a few
others. Add <tt>...</tt>
around pathnames all
over the website, and make all lines (to the maximum extent possible)
shorter than 80 characters. Some ports work maybe (but then, I have got
two people who can do that, since base keeps me busy enough). Install
new HDD into HERC and mop up; backup and set up cleanly ODEM. Some stuff
which I noticed while my system was making a release and I were in the
bed, on the toilet or in the train which I just can't remember now.
Get more sleep. Write a website describing my services (setting up and
maintaining MirOS based communication servers); set up my printer (not
the MP3 one, the Epson FX-80 :). Et cetera.
09.01.2004 by tg@
Add new CVSweb code which is fixed up a little;
add anonRSYNC services at BSDadvocacy (thanks selerius). Later that night,
remove trailing whitespace in this file (*ALERT*), and
wrap the lines at <80 when possible.
11.01.2004 by tg@
Code some cosmetics, fixes and improvements for
RCDB, the issues have been mostly pointed out by bsiegert@ and Waldi.
Move some DIAGNOSTIC messages to DEBUG ones, they were too annoying.
Add proftpd and bsdiff ports from the OpenBSD ports mailing list, where
nobody seemed to care. Install all share/doc/papers/ and have
them build correctly. In dhclient, don't overwrite /etc/resolv.conf
if /etc/resolv.conf.lock exists. Apply wbx@ jumbo patch for the
base system to replace the bcopy, bcmp and bzero functions with their ISO C
counterparts, and fix some bcopy, bzero, sprintf, strcpy, strcat myself.
Make netstart slower by using a fake resolv.conf at set-up time. Add a
webpage describing the various logos. Play a bit in <bsd.port.mk>.
Later, after polishing the website, import GNU librx (LGPL) which is
needed by GNU Pascal's regex unit, and an alpha development shapshot of
GNU pascal itself (the last version is too old for even gcc 3.0); thus
postponing the update to gcc 3.3.3 or 3.4.0 since it's not production
quality yet. Also remove a few unused files. (rx and gpc are not connected
to the build yet.) Disable splitting the texinfo files as new MirOS
standard. Update rcdb, preparing to move it into libc, together with
bsiegert@ the libdb guy ;-)
14.01.2004 by tg@
We will ship MirOS #8 as soon as possible,
and MirOS #9 as the MirOS #8 release was scheduled; probably later.
This is due to the incident with the server checksum failures. The
first MirOS Linux version will then be MirOS X.
Fix remaining
issues with rcdb; move it into libc. ANSIfy anoncvssh. Help midiplay(1)
with .rmi files. Sync ports with OpenBSD (base later).
15.01.2004 by tg@
Continued merging of MirPorts. I feel
slightly overworked.
20.01.2004 by tg@
Finish merging MirPorts; start fixing the
framework as well as some single ports. In the Netzladen, fix rcdb:
finish the man page (proof-read), and fix a major and some minor bug.
Re-decide on the development plan after wbx@ has left the team; we
agreed on not putting strain on ourselfes, but rather prolonging the
existence of #7-stable. The src/distrib/sets and
src/X-Window/distrib directories leave the tree (well, except
for the notes, but these will be taken care of later), in favour of
the new-born src/distrib/lists and generic src/scripts;
there are two new sets: perl and egcs (gcc3, including perl parts of it).
Fix even irssi. Hey, I had a hard day! CSU Switch in
i386: crtbegin.o, crtbeginS.o, crtend.o and crtendS.o are now taken
from the FSF code (licence: GPL with binary library exception, ie. no
practical difference for the most of us) – this might eventually lead
to better Wine support. Updated config.guess and config.sub from the
FSF master; submit local diffs to them upstream. Update lynx to 2.8.5dev.17d,
fixing at least one common crash. Update and ~/.etc/fooify
icewm. Add our PGP public keys to contrib; SSH will follow. Sync egcs
with OpenBSD diffs against FSF (yields a better infrastructure in parts;
the ability to regenerate files with yacc; AMD64 support; the sentinel
attribute; better backwards-compatible (to MirBSD #1? huh?) C++ behaviour.
Fix and improve libstdc++ while here. Also, clean up Pascal.
26.01.2004 by tg@
Bump library versions accordingly and fix building
the base system. Honour ${DESTDIR} correctly in libstdc++ – previously,
includes weren't installed correctly. Remove the need to include a
README-fixinc. Remove all Makefile.am (GNU Automake) and fix a shitload
full of build time warnings. Update config.rpath as well. Add egcs and
perl sets, and merge xshare into xbase; update docs appropiately. Shuffle
order in which sets are installed, move tinyirc back into base. Update
graphics/livestream (distfile vanished). Update ld.so to not issue warnings
about being unable to resolve weak symbols. Whitespace whack on files
touched. Stub out locale support in libc even more. Enable building of and
linking against a shared libgcc (XXX what about static arches? to be
revisited when it should be needed). Transgaming WineX now
works. (To gain even better support, mount procfs to /proc
with -o linux).
31.01.2004 by tg@
Update and/or fix several ports, partially from
outside of MirOS Project (Robert Nagy, mostly). Update CVSweb cgi. Write
a GNU as(1) Intel-syntax howto. Integrate mod-fortune-fvl into base. Sync
with OpenBSD base. Add scripts which help importing stuff. Migrate Ports
infrastructure, and play around another few things – REQUIRE_EMUL is new.
Fix /etc/rc with respect to mounting filesystems. The default
IPv4 route is now asked for per interface, and correctly propagated after
installation. The a.out tools leave the tree. Regenerate kernel files,
and re-think about what is enabled and what not. Keep the small floppy free
of additional stuff like find(1). Fix dict(1) – miranda.org seems down, so
we'd use dict.org instead. Prepare for 64-bit MirOS on x86. Update some of
the propolice stuff, but not everything, to the latest version. Increase
randomness taken at bootup. Move eqn to neqn(1), as it belongs; improve
building of man pages. Cope for new licences (Apache 2.0, XFree86 1.1).
Ensure that ports/packages ISOs are FTP exportable. Sendmail stuff done on
making a release is now done in the sendmail/cf/cf directory and depends on
not being in SKIPDIR (as well as some other stuff). X-Window synched with
OpenBSD; evilwm and lynx with upstream release resp. development snapshot.
Update djbdns for ICANN B root server IPv4 change. Fix building some ports
for ekkoBSD and MirBSD uname types. Greatly improve AnonCVSsh (for more
modularity – eg. on BSDadvocacy -, more flexibility, better structure); all
options are in the Makefile now. Offer bsd.slim on installation, too. Fix a
really deep bug in rcdb for both -current and #7-stable (pkg_install).
Prepare build of a release-quality snapshot in chroot. Work on enabling the
system to build with a read-only ${BSDSRCDIR} (notwithstanding the symlinks
into ${.OBJDIR} or /sys/arch/*/compile). Get flamed by the OpenBSD porters
for this maintainer-variable thing again – I would really like to know where
this will lead us to. If these guys didn't want "to make us work for you",
they should not have published it in the beginning, and set it under the BSD
licence. We all do not understand what they want from us – maybe they just
fear we could take their users away? Probably I should just make that the
project aim.
All announcements etc. will be signed from now.
11.02.2004 by bsiegert@ (catch-up entry)
Fix gnupg (security). Finally install the new harddisk for amiras and re-use
the old one for frankie (alias benz.mine.nu). Just about time, because
the old one from frankie gave some CRC errors while copying :(. amiras now
has 120 GB, frankie 40 GB. Install -current on amiras's wd0d.
Import OpenBSD ports again (twice---I hate CVS) and help with the merge.
Update Berkeley DB to 4.2.52.1 and subversion to 0.37.0. Please test this
release, it might become 1.0. Create a work-around for libtool 1.4 to work
with pgcc and apply it to some of the problematic ports---this needs to be
done further. .etc support for www/snownews. Incorporate CUPS
patches from Jakob Meuser.
After getting flamed by some people (I won't mention names here) over and
over again, remove OpenBSD's port maintainers. This had been forgotten after
the last few imports. Oh, and explain what "vendor branches" are and why
they contain the unmodified OpenBSD source.
15.02.2004 by bsiegert@
Fix lots of ports, mostly MirLibtool stuff: libxml 1 and 2, libaudiofile,
xfmail, etc. All those patch-ltmain_sh files are obsolete now and should
be replaced. And, as a special announcement: Mozilla and Mozilla
Firebird work now. Alas, only on -current, because of libgcc and
crt fixes applied there.
16.02.2004 by tg@
Okay, I admit I've been lazy lately. Or busy.
Anyways, here it is: update developers' keys and the mirex port. Enable
build with read-only ${BSDSRCDIR}. Fixes for #7-stable. More acronyms.
Fix generation of initial host.random file in the old install script.
Add vfork memory sharing ("old style behaviour") kernel config option.
Repair XFree pasting and composing by adding back basic locale crap.
Import GNU libtool and config as vendor branches into contrib and make a
"MirLibtool" which is now used throughout MirPorts. Submit config.guess
and config.sub changes to FSF and get positive feedback. Fix bloatzilla
(well, firebird) building of shared libraries and executables. Get the
OpenBSD IPv6 and shmat fixes. Add back some of the IPV4_MAPPED code (not
functional yet, but dajak has promised me to work on this later). Remove
some dead ones in the known_hosts skeleton. Compiler fixes for the case
if POSIXLY_CORRECT=1 is set in the environment (fixes tiff). Update lynx
to the 2.8.5 release. Regenerate configures in src/gnu/, dupes and unused
files/libraries. Add -4/-6 to rdate(8). Fixes for the MirPorts infrastructure
(even in the comments, so don't tell people I wouldn't write docs). New
rcdb API with major improvements. Several updates to src/scripts/ helpers.
Update cvs to 1.11.12 from 1.11.1p1 and retain local changes. Get new
GNU autoconf ports (XXX they all should get the fixes (eg, for
GNUSYSYEM_AUX_DIR) from 2.13). Update centericq. Fixes for <sys.mk>.
Unbreak imap-uw (and install that on herc). Disable XFT by default and
fix the Standard Editor. Remove ports/infrastructure/patches by converting
them (and a few of the USE_LIBTOOL=Yes ports, too) to MirLibtool which
has been intrigued into MirPorts then (well, except for that it needs a
sync, but this won't be done before the GNU config guys updaye...). Enable
the TrueType bytecode interpreter in the freetype port if not MKC_USAP.
MirLibtoolify a bunch of otherwise harmless ports. MFC back the syslog.conf
fixes for sparc. Add new CONFIGURE_STYLE autogen (XXX only for porters who
know what they're doing!). Decide that libtool 1.5 isn't possibly ready for
inclusion at all. Update propolice from OpenBSD-etoh instead of IBM-etoh;
he finally admits the breakage and works around. Write this entry ;-)
02.03.2004 by tg@
Fix the previous entry's HTML. Merge FSF files in contrib/
and simplify further. Add closefrom to ports systrace infrastructure. Add
PDF printer and enhance MP3 printer in /etc/printcap. Add more binutils-alike
man pages where the newer ones are GFDL'd. Fix a huge bunch of ports and
patch around like crazy in the infrastructure. Add files there: autogen.sh,
config.rpath, mkinstalldirs, move-if-change, missing, install-sh. Switch to
new-style bootsystem MBR which now displays the types of the partitions to
choose from. Get mad on Schily ports again and cope for some breakage of
theirs. Change arch(1) to output MirBSD. Merge ports import from OpenBSD
with bsiegert@ (surely he's done more work than me, who was busy with his
crappy notebook). AUTOCONF_NEW goes to 2.59 and mod_gcc3 bites the dust.
Revert cvs back to 1.11.1p1, because the newer version had core dumps on
joining. "make plist" finally has gotten rid of checking dependant
packages. Someone needs to fix my speling. Add "@" to morse. Fix xlock font
for xjack mode. Update contact info in #7-stable. Remove TCP_LOWDELAY from
sftp(1) connections as well. Let user choose which pager (cat, less, more,
${PAGER}) he wants to use in pkg_add(1) MirBSD pkgtools. Add backwards
formats in cpio(1) mode to pax(1). Add -R and -S options to tar(1) mode for
generating CPIO-format archives. Have MirBSD pkgtools create CPIO instead
of USTAR archives by default; change extension to .cgz appropriately(sic!)
and cope for .cgz etc. distfiles. Fix CPIO and LHArc support in misc/mc.
Enable USE_SYSTRACE by default.
27.02.2004, rotate ChangeLog.
Move some stuff around in GCC configuration which might help cross builds.
Improve operating system checks in MirPorts infrastructure; OpenBSD 3.5
will be added soon. Large cleanup in MirPkgtools, including enable pkg_sign
to build (untested). MFC back latest pkgtools to #7-stable. Convert GNAT
(the GCC 3.x Ada compiler front-end) to use BSD make instead of GNU make.
Have libstdc++, libobjc and the Fortran libs use standard make prereq, make
includes mechanism of installing their headers. Enable the Ada, Objective C
and Fortran 77 front-ends in addition to C and C++ ones. Track down Pascal
until getting to a dead point. Support installing specially generated ports
and source tarballs as sets during the installation process. Enable UTF-8
locale support in X-Window which might somewhat help. Break irssi because
it's unfixable on #7. Initiate building a nonpublic snapshot.
Yes, I do
know that XFree86 4.4 is out. Please give me some time before merging it,
and handle src/X-Window/ as broken in the meanwhile. Arigatou Gosaimasu.
13.03.2004 by tg@
Fix build and installation issues. Import XFree86 4.4.0
and make it work, including the xlock C++ modes. Remove ports' ambitions to
change sysctls automatically. Add new www front page. Show adjtime remainder
in rdate on request. Add new MBR, this time stable, and remove the El Torito
probe (this change adds a requirement to El Torito boot: bootable MirOS CDs
must be generated using mkisofs -no-emul-boot
-boot-load-size 4 -boot-info-table and no other ISO
generation program will work, neither will it work if burnt as is. More
patching in MirPorts infrastructure for autoconf support. Speed up rcdb by
omitting default sync behaviour. Add cksfv port. Enable bytecode
interpreter for TrueType™ files (outside the USA, because Apple has
patents on it). Throw libpng into the base system. Delete redundant copies
of libz in the tree. Install protoize, gcov and the gcc3-cpp if gcc2 is
not installed afterwards. Reduce number of exported symbols in DLLs. Build
the Pascal and Jav^H^H^H"Compiler for a Java™-compatible
language" front-ends and install the Pascal RTS and utils, too (tested).
Fix remote DoS (by OpenBSD). Use CPIO naming conventions for files and clean
up. Update SSH. Add fchroot(2). Add support for ports/INDEX grepping to wtf.
Remove hardcopies of X11 docs. Add back some libstdc++ documentation files
which previously had to be removed because of filename length limit in the
USTAR archive format by POSIX. Enable group wheel to sudo by default. Upload
a new snapshot, announce that.
I lost my laptops' hard disc contents,
reminder: never change the partition table entry pointing to the disklabel
when a slice of it is mounted and active...
14.03.2004 by tg@
It was a wonderful day outside, and I've enjoyed it.
I had to fix a thinko in the sudoers file, and have built numerous ports
again, but still no idea how to recover my dump(8) file or to clean up
my room :) – There is a lot of press coverage recently, and BitTorrent
will make good download mirrors sooner or later. I'm now waiting for my
dedicated server order to be delivered, and will probably install the
snapshot on it. Seems stable enough. I might get a 200-400 MHz i386 donated
in order to build #7-stable and test -current on it. I hope...
There
is a new IPv6 with SixXS and MirOS howto.
17.03.2004 by tg@
Update this file, and the website per se (CSS from
bsiegert was improved, but still had room for further improvements; now it
looks nice on MSIE 5.00/Win2k as well (wow!). As for the rest, I'm busy with
my dayjob, but fond of attention we get (even on the mailing list). The
weather is nice outside, so I'll recover there a bit. I don't have internet
at work now, only proxied IPV4/TCP/www which isn't enough. Sad.
19.03.2004 by tg@
After work, where I learned more about ksh again,
I've started discussing the licence template, and the ekkoBSD people, albeit
still unsure about to use the OpenBSD ports tree, the MirPorts Framework or
the NetBSD pkgsrc, happily joined the discussion (and mailing lists).
I've added a man.cgi to www/ whose only job is to fork to man.sh on all
(official) MirOS website mirrors. Maybe we'll have online manual pages soon,
but I'm not rushing instead of thinking and pondering...
22.03.2004 by tg@
I'm building a new snapshot, and in the midst of
the process, in a sudden, I get the OK for my dedicated server. I'm shocked
by /command, /home/[0-9a-z]/, visas, confixx, SuSE (defective) and friends;
thus, I'm already planning how to update that beast. This basically means that
I've got to throw a ramdisk kernel with serial console onto it, and somehow
chain into the loader from lilo. (Shouldn't be too difficult.) Also, I'd like
to keep this "Leenux" thingie on it for recovery purposes. So I've got to make
the partition smaller. – Anyways, I'll discover something.
28.04.2004 by tg@
Yep, we've been slacking – but it's worth it; while
rsynching a backup of CVS today I noticed it's 1337 MiB in 95842 files.
02.05.2004 by tg@
Tentative update – loosely chronologically sorted, but mainly threaded;
superseded changes haven't been mentioned.
Various reliability fixes (xlock, libssl, ...). Reduce amount of directories
in mtree database, add some entries to the daily job. Fix png in MirPorts
various times; adjust for OpenBSD (3.4, 3.5) and ekkoBSD; both are USA-based
(patent issues). Heavy work on fixing and making mirbsdksh portable; release
some distfiles of it. Fix vnconfig (one could
configure but not reliably unconfigure read-only files). Clean up after
removing xfs, netccitt, netiso, netns, ... and fix Ada. Add netatalk to the
GENERIC kernel. modgcc3 (ports) dies. Modernize djbdns (dnscache) roots.
Change syntax of a bunch of ports variables. Add md4(1). Fix a couple of
ports, update some, make some new. Add cacert.org key to SSL, fix perms of
/var/mail. Finally enable async mounts on ext2fs (from bsieger@).
Manpage fixes. Have ksh check for mkstemp(), strlcpy(), strlcat(),
arc4random() availability. Add some acronymes and sync with NetBSD. Fix and
add no_x11 flavour to Ghostscript. Add pascal and ${fooFLAGS_foo.o} support
to <sys.mk>, and $LINKER support to <bsd.prog.mk>. Remove a
plethora of warnings throughout the tree, and add a new kind of warning: a
link-time warning that linking a program with a certain library (for
example, readline or iberty) will GPL it. Rewrite some scripts using my newly
gained KSH knowledge. Update e2fsprogs (resize2fs tested). Add libtool 1.5
support to MirLibtool. Switch base system from libtool-gcc3 to MirLibtool-1.4
and regenerate stuff. Fix sparcstation 5. Prepare binary MIRBSD_7 (MirOS
#7-stable) updates, and a re-rolled system (MIRBSD_7quater, MirOS #7quater).
Add pkg_scan after suggestion and code from bsiegert@. Brand ftp as MirOS,
finally. Add more entropy to pgp-2.6.3in port. Add CSS selection support
to this homepage. Update CVS after security flaws, and pull in support for
better expanding of the local tag from CVS-stable. Finally matroska/MKV
support for mplayer – feeded to OpenBSD, but no interest from them. Remove
a lot of unused files (local md5, snprintf, zlib copies). Update protector.c
from OpenBSD (hard work, to do it manually until it breaks no more). Help to
un-GPL the gpc RTS (hey, there's still no gcjrts port!). Remove useradd,
userdel, adduser, rmuser, addgroup, rmgroup, sysctl writes and more stuff like
that from all ports. Rewrite strlcpy and strlcat, speeds up; switch to -O2
for gcc3, too. Improve randomness collection – caveat (from Ignatios
Souvatzis): it might reduce entropy level, thus disablable with a sysctl.
Fix the installer. Add support to MBR and fdisk(8) to force loading an entry
which is not 0x80 by default (application sample: NT chokes to load if it's
not 0x80, and there's another primary, DOS, partition which comes before the
NT partition in the partition table -> set NT to 0x80 and fdef the
BSD partition which I want booted by Return or the 10-second delay). Optimize
generation and installation of the GPC RTS and split out the utils into
gnu/usr.bin/ separately. Update Lynx to current development snapshot. Add
handler for .cgz/.ngz -> .cpio in gunzip(1). Add -femit-memmove to both gccs
and document that; add capability (HAS_GCC2 / HAS_GCC3) handling; use that
for descending, libgcc.so, cpp installation; adjust manpages and their
installation routine accordingly. Fix ${SHELL} in Makefiles. Get reliability
fix (SHM) for -current, sync contrib/gnu files, users/groups, and start
updating this website. Get OK for torrents. Disabled fortran libraries (libg2c
and libfrtbegin) because of unknown breakage, temporarily.
04.05.2004 by tg@
Fix a bunch of issues documented on the new
errata page, make another snapshot, release that via
BitTorrent (thanks to Eike Frost, Scarywater (Anime ;) Tracker admin).
Go over the website once again. Fix the libpng issue post-snapshot.
Man am I tired, I should go to bed (I've got a dayjob, y'know?).
07.05.2004 by tg@
Make website more legible (smaller font); revive
the old (works well in konqueror, bloatzilla is broken) CSS. Incorporate
hints (documentation-wise). Update libebml, libmatroska, livestream to
new versions and after that, mplayer (depending on these). Add port for
gentium font; ask for licence clarifications. Fix MBR timeout.
08.05.2004 by tg@
Some sparc make-it-easy things, although the
whole sparc/stand/* stuff smells funny as well. Simplify gcc config.
Add <stdint.h> header file (C99 required); fixup some
other headers (and, in consequence, apps which are confused that 0
!= NULL *sigh*). Prevent stuff from being placed into
the source tree at compile-time; start merging OpenBSD: don't place
OpenBSD "this file is not ELF" markers any more; recognize MirOS BSD
PT_NOTE ELF sections (currently, MirBSD and OpenBSD are still handled
the same), and feed fixes to ekkoBSD while here. Try to fix random
(hah! ambiguous) hangs, and fixup timeout_set() while here.
Some minor fixes in the build tree (NOMAN in distrib/special, CFLAGS
-> CPPFLAGS, etc.) and dependencies. Replace misuse of sh, ksh,
/bin/sh, /bin/ksh etc. in some Makefiles by ${SHELL} – there's more
to do. Remove "SLIM" kernel. Switch to MirOS BSD PT_NOTE ELF section
and clean up src/lib/csu/ while there. Release "corrigendum"
torrent for users of the old torrent snapshot.
09.05.2004 by tg@
Add protest action call against software
patents to redirector and home page. Announce the update package.
Fix up some ports. Add wiki.mirbsd.de with help of waldi
(Bastian Blank) and a UseModWiki to the www module (still experimental);
code up a lot of things for it (ISO 646; config options; look and
feel á la MirOS; force https; display User ID with name;
add login button; encrypt passwords (using old Unix® crypt
hash – UNIX® is a trademark of about fourty different organisations
over the last thirty years, so heck I've given up); fixup whitespace,
syntax). Fix some warnings in the PHP part of the website too, and
call for discussion on simplifying (mostly) the MirOS licence template
(deadline is Wednesday noon; after that, I'll commit the now-draft and
set up the Wiki). Even have Wiki config files in CVS, and hide Wiki and
CVSweb configs from Apache. Watch anime ;-)
09.05.2004 by tg@
I'm backing up and reinstalling the main
project server, HERC, with the latest snapshot. That's why services
can be a bit unavailable right now.
15.05.2004 by tg@
I've fixed a couple of things while updating
the server and added others, for example the new dynip scripts in
the codesnippets section at contrib. Pine got updated, and bmf and
sendmail play nicer together; some kernel random bugs are fixed again.
Also mailx(1) works with empty $SHELL now, and mpg123 compiles without
warnings. Some more YP tools died. We got licence clarifications from
the gentium font creator, and clarified our own licence; this triggered
enabling the [link] MirWiki after all.
Maybe, anonRSYNC already works again; and we're at gmane.org now.
19.05.2004 by tg@
Yes I know I should merge that import, but
hey, at least we've got a secured-again cvs now. (No need to hurry to
update the anoncvs-bins, but anything not chrooted, not anonymous
should be updated.) – As for my private life, I'll be taking a break
now, but not without having bunkered enough (100 bottles) of Club-Mate and IceT at home ;-)
21.05.2004 by bsiegert@
Okay, I have been writing nothing here for several months, but there are
some Organic Chemistry courses which are keeping me incredibly busy. I
work on MirOS when I have some free time. Right now, I am merging all
the non-security MFCs from OpenBSD 3.4-stable for #7quater which should
be released RSN. I have also done some work on MirPorts: import some
updated ports from OpenBSD (all security-related stuff), add a
chemistry category with only one port (chemtool), fix some
stuff for #7-stable and -current. Add MirOS snapshot announcements at
freshmeat.
Hmm, I should really write these entries more often...
22.05.2004 by tg@
It's been good weather again, thus I went for
a hike with a friend two days ago, and made some potatoe salad myself
(instead of buying) – you know, cooking is a good diversion to do when
bored of coding and hanging around in the Netzladen, and some people
say cooking is the same thing as doing chemistry ;)
I've finally
started to merge over some of the OpenBSD stuff, and found just that
my TODO is increasing
instead of decreasing – for example, I'll have to rewrite the installer
(I've got to do it anyways, but now's the time, because if I didn't, I
had to do major changes in the existing code anyways). We've discovered
that libtool 1.5 doesn't work without a C++ preprocessor/compiler on
board which plainly sucks. I've allowed admins and editors to upload
files to the MirWiki, so we can present some small screenshots – there
are some of my evilwm environment already. Merging was done in regress,
usr.sbin, games and distrib; the OpenBSD Perl package tools left the
tree as well as makefs(8) which I never got working (for the same reason
as the old SPARCstation 20 with the 50 MHz CPU I was having which we
couldn't debug otherwise than to get a different, 75 MHz CPU for it).
The usr.sbin/pkg directory left and went into pkg_install/pkg (I'd
rather have it done the other way round, but CVS doesn't allow to move
files or directories, or even track imports after that, so it'd been
a PITA). Also the includes directory has been merged or adapted to be
compatible with new code, and now does indeed work without GCC3. I'll
add a non-gcc3 source checkout to the modules file. The crunch tools
went into usr.bin and – what irony – are a crunched binary now, also
I fixed some stuff and added some minor paranoia error checking to
the code. And I updated the MirOS CA CRL (three days late, but hey),
with the HERC server certificate revoked – it's using CAcert.org as
an experimental test phase now. I'll head for breakfast now, and be
back to code/merge later.
02.06.2004 by tg@
Our beloved event LinuxTag
is coming near, and I've ordered some BSD Dæmon pins. We will
also (probably) be there all days and sell limited MirOS T-Shirts.
Well, what have we done? The usual stabilizing, of course – even the
infrastructure (-p to rsync). Fixes, MFCs, optimizations. I've started
to merge OpenBSD 3.5-current (not yet finished: src/gnu/ and a few
ports). We've got a MirOS PT_NOTE ELF section scheme now, and the
kernel and file(1) already recognize it for BSD and even Linux. I've
revived the old usr.bin/learn program and released mirbsdksh-1.10; also,
another licence discussion happened, with cool feedback from the
ekkoBSD people. The kernel should
have shrunk a bit by now, at least the ramdisk kernel, by leaving out
unnecessary OpenBSD 2.1 compatibility (we'll have a COMPAT_OPENBSD
later, and most ancient subroutines were used by other emuls anyways).
I've played around with Interix/Microsoft® Services for Unix® 3.5
a bit, before and during a long LAN party (with Windows® games, yea
right) this weekend – and I think, once I discover if Interix has shared
objects, and once MirOS Linux works, I might hack a MirOS Interix.
I've converted more and more scripts to mirbsdksh, and tested the latter
under Interix, Mac OSX, GNU and Solaris, of course. Also, there was a
small start of a licence mop, and an XF4 merge from OpenBSD. Clean up
the /etc/profile file (works as ~/.profile under my
Windows®/Interix environment *g*). Attempt merging more
ports, and other stuff like that, so bsiegert@ is happy. Prepare for the
upcoming MirOS #7quater release (this time for real).
04.06.2004 by bsiegert@
My Organic Chemistry course is over now, though the next exam is not
far away. Import and merge OpenBSD ports, leaving some bits and the
infrastructure to tg. Add the last OpenBSD 3.4-stable MFCs and security
updates and build a preliminary release. Discover thar jpeg does not build
and fix the build system---by rewriting everything using
bsd.{prog,lib,subdir}.mk :). It works fine, and it is really
elegant. Try to do a bulk build of all the packages for #7quater.
04.06.2004 by tg@
We're apparently just two steps far from the
#7quater release (a.k.a. "It's still #7, just that the binaries
are uploaded for the fourth time!"). That would be, waiting for
my slow SPARCstation 20 to finish, and fixing (ugh!) Perl ports.
05.06.2004 by tg@
As for Interix – I've discovered one can
build shared libraries, and they use OpenBSD's ld.so... now I just need
to prod them into giving the sources to us (maybe even under a non-OSD
compliant licence which would still qualify it living in contrib/ for
the sake of porting MirOS to the Interix/NT kernel duo). But that's
just dreams for a future far away.
In the current, we (bsiegert@
and yours truly) are just building MirOS #7quater, whose source was
finished yesterday (indeed, it seems to build for me, while bsiegert@
has some weird errors... which I cannot reproduce. In MirOS-current,
I've fixed a shitload of bugs and malfunction again (in fact, you do
not want to use even the latest snapshot, but you cannot build a more
recent source tree. Fucking situation)). I will now import OpenBSD
again and merge src/gnu/ as soon as possible, then jump to -current
myself, remove kludges and merge the remainder of the new OpenBSD
import. I will then stabilize the shit and publish another snapshot.
And I'll wait for ekkoBSD to import OpenBSD sources, merge their
src/gnu/usr.bin/binutils (so you can cross-build from MirOS to ekkoBSD,
if you are so brave) and split out "our" configs. Then I'll have a
look at making the target architecture suppliable by the user, maybe
move more stuff to <sys.mk> and so prepare for a (future) cross
building (I might try to build sparc cross, and I will definitively
bootstrap MirOS Linux cross). Also, I ought to really check if the
whole system builds through without gcc3, without shared libs, etc.
09.06.2004 by tg@
Fixing the release and newfs-current (I seem
to have broken mfs). Working with time structures and leap seconds
again, is MirOS now an unPOSIXly OS which shits on XPG4 and produces
the correct time – even inventing a new API. I'll probably go 64 bit
time_t RSN – and definitively throw out SMP (agreed bsiegert@). Also
I got to fix div.S – there may be NetBSD guys who can't do assembly
right (no wonder, they were using AT&T syntax). And OpenBSD guys
who don't notice. Update ports infrastructure, and import src/gnu/
for merging – discover CVS bigfix while doing that. Hunger.
10.06.2004 by bsiegert@
I think I will give up the bulk build. It just takes too much time.
frankie's NIC (a "Compaq Embedded Netelligent 10/100 TX") seems to
be faulty, add a fxp ("Intel EtherExpress Pro/100S") -- works
like a charm. Install MirOS BSD #7quater (not yet generally released)
on frankie. The official release is going to be at LinuxTag. Debug cups
with help from Bastian "waldi" Blank (local LUG meeting) and tg@: It
actually triggers a bug in the low-level assembler parts of
libc. tg has rewritten div.S. The result is that cups
will work in the next MirOS BSD snapshot. Plug MirPorts again at
ports@.
11.06.2004 by tg@
I've merged OpenBSD's src/gnu/ finally, fixed
the binutils stuff (hello OSABI, PT_NOTE etc.) and contacted ekkoBSD
about this. (I might just add support for their OS, always assuming
they don't differ from OpenBSD, but don't do that yet because they
_might_ have done it already, differently or something.) I also had to
account for the PCC_STRUCT_RETURN of MirOS (like NetBSD) vs REG_STRUCT_RETURN
of OpenBSD. I've fixed more timing stuff, indicated the correct epoch and
counting method in certain manpages and use now time_t as element of the
timeval struct. I've fixed binutils that much it built (at least) once with
-Werror, cared about dynamically generated content and fixed Makefiles.
I found the culprit who set SHELL= (empty) in my environment – it
was GNU screen's config file, and fixed that in the port. I've made RAMDISK,
the former CDROM kernel, look like GENERIC and put ssh/sshd on it (untested);
then fixed building the kernel and base system up to (and including) isdnd;
including, but not limited to, API clarifications, NULL vs 0 vs '\0' issues,
further string cleanup in AT&T code and deuglification. Lint is a lot
more quiet now, and our web server is now called mhttpd.
15.06.2004 by tg@
Okay, let's go to bed and become well again. I've
finally got myself around to bring MirOS #7quater on the server, the release
on an ISO, that on CD and basically tested, and the T-Shirts to the copy
shop. I've done a few ports and ports fixes at Kabelaffe's, then did a full
build on mine. During that, I fixed a plethora of warnings and invented some
fun new stuff, and made crap adhere to more modern standards – which led me
into polishing and rewriting most of the infrastructure. We've also got a
new attempt at an Ethertap device (tun device with link0 set) by Claudio
Jeker, although untested, and an external set of patches and stuff for
MirPorts on OpenBSD and ekkoBSD. Benny has written a release announcement
(mostly copy and paste tho), but since -stable is his, it's alright. I
first thought a <bsd.wrapcf.mk> for all these Makefile.bsd-wrapper
would be cool, but found it undoable (but may look after it later again).
16.06.2004 by tg@
I've ordered 40 printed T-Shirts, started converting
stuff to <bsd.cfwrap.mk> (it's not impossible, just very hard)
and fixed a few minor things in the kernel – among others, you can set the
"uname -s" for native Linux binaries to "MirLinux" now;
believe me, that will help me bootstrapping the whole shit.
17.06.2004 by tg@
Okay, after more infrastructural work I've gotten
around and publishd a snapshot. If you use IPv4 default routes during
install, you will have to fix it manually. If it does not boot, the
bootloader is broken – I did not test it. It won't execute old binaries,
and the next snapshot I will publish cannot execute the binaries built
under this snapshot. It's merely a for-fun release.
20.6.2004 by bsiegert@
Create from scratch and commit a port for ESP Ghostscript. This is to be the
first step for a proper printing support architecture, with all kinds of
printer drivers. The next step will be gimp-print support in espgs and
CUPS.
Create a flyer
(actually a leporello, to be folded twice) for LinuxTag 2004. My cousin
has created a huge (91x85 cm) poster (a variation of the original
"logo" by insk) in CorelDRAW. I will try to have it printed at the university
on Monday. I still have an exam to do on Monday before I can go to the
LinuxTag. Wish me luck.
21.6.2004 by bsiegert@
Okay, I have collected all the MirOS paraphernalia for the LinuxTag. The poster
is printed, I have copied 100 flyers and even made some name tags for all the
booth slaves. Tomorrow, we will decorate our booths and fetch the exhibitor's
passes.
24.6.2004 by bsiegert@ (LinuxTag 2004)
Yesterday was fun. We sold quite a few T-Shirts (although not as many as we
had hoped), spent some time hacking, and had a few interesting talks. I unbroke
the cross-platform make Makefile.boot for use on Darwin and did some
MirLibtool thingies. Till Kamppeter from
linuxprinting.org told me how to
do foomatic/Ghostscript printer drivers right — I will try to make a
port with this information.
The nice people at the Sharp booth allowed us to print out some flyers for
our project. Over the day, I brought ten or so PS files, but all of them failed
to get the Gentium-Italic font right. Today I am going to try a PDF file...
25.6.2004 by bsiegert@ (LinuxTag 2004)
For the record, here is what did the trick with the PS file: Add all the fonts
to Ghostscript's Fontmap and convert the file to PDF and back to PS.
Install OpenBSD-current on manitou (iBook) and actually test the mpsup stuff.
Fix some flaws, especially with the __RCSID() macro. This macro
should not be used in MirPorts for portability reasons. Update vim. On the
booth, I am always asked the same questions: "What is the difference to
OpenBSD?", "Why another fork?" etc. These questions are of course answered
thoroughly in our
flyer.
29.06.2004 by tg@
Okay, LinuxTag is over, and the new torrents are
made available. I'm still trying to catch up, and I'll likely install MirOS
on my shiny
new Soekris
box these days. Maybe I'll fix all the other staff first, though.
Alright, joe does utf-8 now, and does it well. There's a new cvsroot up.
08.7.2004 by bsiegert@
Fix lots of ports—mostly those who showed up during a "make package"
in the x11 directory. Fluxbox assumes that time_t and int are of the
same length and thus
fails
miserably. I won't work on GNOME until the next
import which is when we will have 2.6. teTeX ships with broken autoconf
stuff, so it is broken too, alas.
I will be on vacation from tomorrow to next Tuesday, so be patient when
contacting me.
12.07.2004 by tg@
Found some errors in the IPv6
page when helping Kabelaffe to fix up his stuff. Besides that, I'm
trying to recover from a light influenza, bought an external 3.5" USB
IDE HDD for odem (still looking for a _new_ cheap 2.5" IDE HDD for the
soekris...), and will release a new snapshot laters. I should fix some
more stuff before, but... I just can't find the time for hacking right
now, maybe I should get myself a break, just like our friends at ekkoBSD
recently did.
I'm wondering, why we don't do the following: first
sync src with openbsd, nuke SMP, fix stuff (especially installer and
bootloader) as well as possible. After the system works at all (ie.
while nuking SMP), import openbsd ports, and fix them all, no
kidding. Then, we'll look where to import or update (I'd like to bump
KDE first, and Benny might want to have the most current GNU GNOME)
or fix, and push out #8 as interim release, before starting with new
things (maybe even MirOS X?).
20.07.2004 by tg@
We've got a new developer, and much work to
do. For a better overview, read the miros-discuss@ and miros-changes@
archives at gmane.os.miros.*, and/or
the Changelogs. I'm also pretty busy with my dayjob and other daily
life at the moment, and frustrated from accidentally deleting the
nearly-finished 10hr effort of merging and enhancing httpd.
22.07.2004 by tg@
Read the gmane archives, it's too much.
30.07.2004 by bsiegert@
Our IRC channel, #deutsch, has moved from freenode to OFTC. We are
going to have a channel meeting (IRL) this weekend at Cologne.
tg@ has written a changelist for the latest snapshot, 200407:
- lots of MirPorts infrastructural fixes
- even more lots of fixes for actual ports
- MirPorts runs on OpenBSD now, does not impact OpenBSD ports tree operation
- MirPorts now installs all subpackages by default
- add trademark appreciation to NetBSD(TM), Mozilla(TM), Firefox(TM), Apache(TM) or remove them where impossible
- MirPorts can use BSD diff now, no need for gdiff any more
- MirPorts package tools string out of bounds access fixes
- switch time_t to 64 bits finally
- gcc3 fixes for format string warnings: the %lld and %b kprintf specifier code was so bad that it was redone
- format string fixes for all of kernel and nearly all of userland (not only time_t-64 related)
- add leap second correction and time_t/mjd/tai64_t conversion code to kernel, fixing CMOS time
- enable pxeboot to build, and install it
- security fixes for PNG and others
- add <bsd.cfwrap.mk> for easier creation and maintenance of GNU autoconf'd software (gcc, rcs, cvs, binutils, ...)
- add default malloc.conf debugging options
10.08.2004 by tg@
Now would be a good time to start doing the
weblog again, without back-catching up, but it's simply too hot around
here in Germany, so I can't get myself up. If you want to track MirOS,
please please read the CVS list. If you really want to track us, join
the IRC channel at irc6.oftc.net:6667 #deutsch and watch the
developers quarreling.
10.08.2004 by tg@
Fixing some leftovers from the libtool and
rest changes. Jointly with sap@ deliver fine online man pages to you,
such as KNF – style(9).
18.08.2004 by tg@
Fix some minor stuff after the snapshot, and
hack libtool to death (not mine, I assure you ;). I'm preparing to send
my laptop to maintenance, so development has slowed down.
28.08.2004 by tg@
Think about a way to do crypto (PGP replacement)
with in-base tools (namely: ksh(1) and openssl(1), openssltool(1)).
This could be used for verifying
signatures for releases or snapshots, rewriting pkg_sign(1) and exchanging signed and/or
encrypted eMail with the MirOS Project officials. Validation expands to
official statements and verification of (other) X.509 certificates, such
as the CAcert.org root cert we're using, or our web server certs. On the
other hand, we can't take liability for stuff signed with the tool...
16.09.2004 by tg@
Benny arrived in France and hacks together code
for the package tools as well as better installation documentation. In the
meanwhile, Stephen starts hacking on MirPorts, and I'm continuously
improving the base system. Oh yes, and a certain, modern, hated-by-me,
webbrowser works (I could use it to view my apprenticeship-school timetable
via ssh-X-forwarding to an OpenBSD/i386/MirPorts box by Waldemar, who's
slacking right now (j/k, he's preparing his promovation or whatnot).).
17.09.2004 by tg@
This all is continuously overloading me; my
dayjob adds to the strain on my body and soul. I'll probably seriously
move the webpages to the Wiki, and not care about it any longer.
I've received copyright transfer contract papers from the FSF and will
send them back signed, so MirOS improvements can get back into GNU.
We'll set up a new CVS root some time in the future, be careful.
25.09.2004 by tg@
Status update. Benny is working hard on package
tools and documentation; I'm currently busy with the two new boxen, loki
and thor. wbx@ helped me to set up LDAP and stuff on loki, so MirSolutions
has got a central user management now.
26.09.2004 by bsiegert
Student life in France is more or less a
continuous party. Alas, I still have no internet in my dorm room, and all
the computers in school have a French keyboard. Thus, frankie is going to
be off line for some more weeks.
In spare time between classes and parties (j/k), I work on completing the
pkgtools. They should support all those new OpenBSD keywords (like
@conflict) soon. Ports are more difficult ATM, but at least i have
completed fire*** 0.8 (1.0PR might come next week).
IMHO, the future of MirPorts will be with our pkgtools, not the perlish ones.
However, PLISTs from OpenBSD should work (mostly) without modification. New
ports however will not use all those fluffy new keywords. Due to internal
differences, packages created with the OpenBSD tools cannot be installed with
our tools. pkg_create(1) will
be able to replace (upgrade) installed packages in the future.
26.09.2004 by tg@
Before starting with the real stuff (porting gcc 3.4
and CVS 1.12), I've played around a bit with OpenNTPD: it lacked (obviously)
correct leap second handling. Doing that, I discovered I had still not
written manual pages for the taitime(3) suite
of functions, and done that as well. And I've added tyler@ to the list of
developers; he can already commit in www/ and will soon be making
improvements to MirOS. My infrastructure starts improving, but all the recent
moves imply some problems to solve, so I'm progressing rather slowly. gecko2
from IRC will visit me IRL tomorrow, so don't expect my pace to increase.
28.09.2004 by tg@
Life goes on... but it's fun sometimes. I gave Dave
and Tyler something to play and imported libcdk, enabled fdialog... have fun.
Tyler tries hard to do a make build release of MirOS, but somehow fails. I
should get him to lend me a GNU screen session on that box, to debug, but he's
always asleep or at classes/work when I'm awake, and vice versa... now what
was next? Oh, gcc 3.4. Won't be the renamed-to-4.0 tree-ssa beast soon.
I could need some sponsoring – I've got a Strato server (again) which costs
39 € per month, and I'll be using it 95% for MirOS and 4% to give out
shell accounts to friends or BSD developers. You might be able to send money
to me via Paypal (I didn't test that yet) or just German bank transfer. Just
drop a mail if interested.
Tyler
has mirrored the website, so add that to the list which already includes advo
and Eimanns etherkiller. Tyler should learn how to write wlog entries :)
30.09.2004 by tg@
Starting the wlog tradition again, or so. I've
started on gcc3 by importing 3.4.2, making a gcc and a coregcc module (not
yet part of MirOS, for © reasons), and threw propolice on that. Nothing
more so far, since I've got a visitor, a dayjob and caught a (very light)
influenza yesterday (probably from my visitor). I'm probably going to set up
thor and fix herc before doing any more work on it, so don't even hold your
breath for it, or C++ support...
Since the others are lazy bum bum
slackers (M-x theo-mode), I'll report a bit on them: bsiegert@ has arrived
in france finally – he's found his papers he needs to order xDSL internet
access (in french). Let's hope he'll be back and go crazy on package tools
soonish. cnuke@ has moved out of his grandmother's house, together with his
girlfriend, and already ordered ADSL; should be there mid of the month. No
answer as to when he'll start committing ports (or www?) fixes, though.
dave@ seems to be alive; tyler@ had a conversation with him. sap@ looked in
into IRC shortly, but doesn't say much these days – he's got a family and MTA
to take care of, pretty reasonable. tyler@ started mirroring the website and
CVS on a university pipe in the USA. (I've gotta do a US-crippled snapshot
and release, probably.) wbx@ still doesn't admit publically that he's a MirOS
developer (again), but working on stuff – both MirOS and vintage related; his
/cvs, codenamed WbxBSD by me, is an OpenBSD/macppc running on an old-world G3
notebook (I had to help him to make the kernel boot, though; even if I hate
Macs). I think after the first MirHackathon he'll become a wee bit more active
on the MirOS core business. Nobody seems to care about MirbsdKsh, MirMake, the
website or MirWiki, though.
MirPG entered the next pre-planning stage;
this is, I'm starting to design (in mind). After that, the first planning
stage will be: writing a positional and design paper (both from a user's POV
and the technical side) in nroff, and mailing that round (first to devel@
then to miros-discuss@), requesting comments. When that's finished, I might
start coding, if my TODO wasn't that big. But there's a fair number of hard,
promising projects in it, and the list is pretty realistic (feature-wise),
at least with regards to the next few things I'll be starting. Oh, by the
way, after the first OpenBSD sync we'll switch to a new, fresh /cvs (I've
already added a bunch of directories to the being-ignored list for imports
from OpenBSD) using 1.12 as the server, and 2.x revision numbers (you might
have to do a fresh checkout, though) and MirOS #8-Beta. — That should
be enough for today, have a nice weekend!
wbx@ still doesn't admit publically that he's a MirOS developer.
08.10.2004 by tg@
I'm recovering from an influenza which keeps me
from having to go to work this week (but also from concentrating on anything,
which is not so good). We should have basic support for i386, sparc and ppc
in our in-tree binutils – thus cross-compiling might be possible at
some time far in the future. (Since my i386 book just died on me today again
for some unknown reason, I don't want my sparcstation to get dusty.)
My to-do-before-gcc-3.4 list has shrunk well enough so that you might be
able to compile your damned C++ code again RSN. But I think I'll fix herc
first, because not being able to watch anime is worse.
tyler@ is a nice
guy – he doesn't write entries here, but starts mirroring stuff, and even
backing that up (which I can't do, mostly because I don't know where to
store the backups, except 5.25" floppies...).
The month has started
quite badly for me: I'm at more than 200 € debt now... and the second
series of T-Shirts is already ordered, will arrive soonish. So I'm open for
donations, if anyone might. And I'm selling my car. Life feels so...
08.10.2004 by bsiegert@
Okay, I haven't really done any MirOS work this week, as I was challenged with
... the real life. Thus, I am going to tell you about the WEI
(week-end d'intégration) which took place last weekend. After
a "fake" inorganic chemistry lecture (a mean trick they play at the
first-years) we headed for a camp in the Vosges mountains on Friday.
The two slogans for the weekend were "dormir, c'est mourir" (to sleep
means to die) and "manger, c'est tricher" (to eat means to cheat). After
dinner, there were some "integration games" modeled after the Herculean
"Dodekathlos", many of which involved drinking. The evening was finished with
loud music and even more alcohol.
That night, nobody got much sleep, because there were patrols who woke up
the people in the sleeping rooms every 30 minutes. The next day, the games
were continued outside. Once again, nothing too bad or disgusting—apart
from one game where you needed to fetch a spoon from a bowl filled with milk
using only your mouth, then doing the same in a bowl with flour (but I didn't
have to do that). On Sunday morning, there was a very nice
randonnée (hike) into the mountains. Most of the rest of Sunday
was spent waiting for the bus which didn't arrive until late afternoon.
So, in summary, we spent a happy, drunk weekend with less than four hours of
sleep per night. I also discovered that I don't like the French Pastis,
which is some sort of anis-flavoured liquor mixed with water.
09.10.2004 by tg@
I'm uploading a snapshot again, we can crossbuild
and tyler has just volunteered to transfer the website to the MirWiki ;-)
My bronchitis is fading away, luckily.
10.10.2004 by tg@
wbx@ leaves for his diploma. I've played 10 to 15
minutes of soccer today. The package tools will leave src/ soon. Nothing
much, but that's today's news.
16.10.2004 by tg@
Benny is lazy, and I've succeeded in building a snapshot for #8-beta
which actually makes me happy. I've ported k7-powernow from OpenBSD-current
with kludges for broken BIOSes, actualised many components of the system
(such as ntpd, grep, diff, sudo, newfs, fdisk, disklabel, mc, parts of libc,
ld.so, libtool, ld, kernel stuff, make) and brougt libexpat into base. I've
done advocacy so the author of the xmlwf(1)
manual page changed the licence away from GNU FDL.
On the other hand,
I've still not gotten around to set up thor completely, and herc's not
powered on either at the moment (but will be testbed for the snapshot).
My living room looks well, compared to a few weeks ago, though. sap@ and
tyler@ are engaged in the Secret T-Shirt For Theo project; we're currently
investigating. The FSF has accepted my copyright assignment forms except
for gcc, where was a typo (d'oh). Well, I've started to assign my src/gnu/
and contrib/gnu/ diffs to them, and on the other side are penetrating OSI
to get the MirOS licence template approved.
25.10.2004 by tg@
I've uploaded another snapshot,
built over night. It contains synchronisation against OpenBSD in a few
core points libraries, programs (e.g. sendmail, httpd, perl, bc,
dc usw.), XFree86™ as well as the latest libpng and libexpat
versions in-tree, and ports fixes (not in the ports8.ngz tarball, but close
enough for a quick CVS update). There have been quite a few fixes in the
system (kernel, ntpd, /etc, build infrastructure, HTML manpages, Perl) too.
I've had quite a fun Starcraft LAN at a friend, two times, this
weekend, and – quel surprise – yesterday evening, a loose friend
appeared at my home and asked if he could sleep here... stressed by his gf
and stuff. Well, I suppose I'm a person you can do this kind of stuff with
;-) I've put him quiet with 2 MiB of ancient DOS games.
Even my
room's quite cleaned up, so I'm げんき (genki), except
for the usual little quarrel with my (currently down) bronchitis.
26.10.2004 by tg@
Work took long today, but went over surprisingly
quick, taking the meeting (which we luckily denote with a German word, not
a denglish one) into account (admittedly, I was next to sleeping in it a
few times, and could barely hold my eyes open). The evening, back home, I
took a 10 minute round ride on my bike to breathe some fresh air, then
did some room cleaning and hanging up clothes from the machine. After that,
I hacked a bit on compat_openbsd(8) until I got to a dead point, then
continued my partial-openbsd-merge work, enabling a threaded perl again.
Unluckily, the OpenBSD design error in getserv*_r(3) functions means a libc
major bump for us — I have not done the errno(3)-caused flag day
though. Miod told me about an easter-egg in the Wiki (hu?), but I talked
him into at least telling me he can't help me this late. Maybe he will help
me hack it together at Fosdem or so. Benny still hangs around in france,
not doing much except cursing about the various TV applications. I despise
television (see my orkut community for that ;).
We've got a special guest tonight.
28.10.2004 by tg@
I'm currently writing this at my dayjob; I've been
sleeping not too much nor too well so I can't concentrate on hacking PHP
with Empress on SuSE GNU/Linux 6.4/7.3 anyway... there is a forum for Germans learning
Japanese, and I linked forth and back, in an example stating how to
embed Kana into HTML ;-) although I haven't quite got the time to learn
Japanese by heart, I'm digging into it, and paving the way for people
who would like to do so with MirOS (the fora are quite Windows®-based,
I've seen only one GNU/Linux post). Marion Zimmer Bradley, the deceased
Fiction author, has brought out a new, final sequel (well, it tells about the
time before the happenings) to the Avalon saga which I've been
literally eating up, chapter-wise, yesterday. Éric
Lévénez has noticed, probably by OSnews, that
MirOS #8-beta (which I still consider "just another development
snapshot" for some unknown reason – the milestone was
a bit earlier, and the next will be after gcc3.4) has been out, and
I've sent him a
few corrections regarding inheritance, stable branches etc. This evening,
I'll be reachable at Netzladen as
usual, since our group's merry meet evening
happens to be, interflicting with a CCC presentation about $otherOS. Wonder
where I'll go? The day after, there's a StarCraft LAN party planned which
I attend, although with a bit of bad faith, because Blizzard/Vivendi/Microsoft
paid a USA court to rule bnetd illegal. (I was going to link to the bnetd
homepage, but looks like they even got that domain! Shame on the land of the
unlimited impossibilities.) Oh, and I've hacked up cgiirc for even better
utf-8 support (Japanese with charmap.exe is NOT fun!). Our frenchman
appears to not do much more than hanging around in IRC and wondering that
I can actually spell français correctly, but I know he's working on
MirPorts in the background. He looks sad since GNU libobjc doesn't offer
the Steeeeeeeeeeve! Mac feeling, and I hope he gets his mind right
on wheter I have got to include it with base or not – I'd prefer to
not, leaving only C for the base system. TCC by Fabrice Bellard now uses
its 8-times-faster-than-gcc speed to compile a Linux kernel as bootloader,
directly booting into the generated object code. Weirdos, but looks fun.
29.10.2004 by tg@
Yesterday, at Netzladen, I've hacked more on the
compatibility modules: time_t 32/64 bit issues. Even the linuxulator was
missing some, and the OpenBSDulator even more. It works now until to the
point where stat(2)/fstat(2)/lstat(2) are called. The previous bug was due
to an oversight in i386/trap.c *sigh*. I discovered how to finally block
the old, 4.3BSD-compat, syscalls in native mode, must look deeper into this;
couldn't have done that since I got to dayjob today again, whilst all the
school children have vacations.
Today, I ported the light keyboard layout
(~/.Xmodmap) of mine, perusing the right Alt key as Meta, to a certain other OS,
and plan to do that for the full layout (with Katakana) as well, if it is
possible.
MirCGIIRC is now fully 8-bit Unicode compliant, and I urge
everybody to use ours instead
of the original CGI:IRC.
31.10.2004 by tg@
Samhain consists of hacking on syscalls, openbsd
compat and gcc 3.4 for me. Still nothing much happening here, but the beer
is good, and mixed with Pepsi. Someone ought to give us a real website, or
better, move all the content into the Wiki. I need a spare machine, an AMD
K6-3/450 or something similar, with 256 MiB RAM, as silent as possible, to
test new packages (on MirOS and OpenBSD) and snapshots I do.
UPDATE:
I didn't get my aching head to hack gcc, but mksh now does spawnvt function
if asked to do so. Man I wish I had gdb to debug these friskin' core dumps.
And kettenis@ and drahn@ still haven't answered whether they have already
set up © assignment papers with the FSF.
01.11.2004 by tg@
Huh, we're in November already? etoh@ told me that
the FSF is slow in processing his papers, but the GCC developers are
interested in SSP, and – although it must be designed differently, for
tree-ssa and other changes – it will probably be in stock gcc 4.1, so
the outside world has noticed "us". Nice. I'm trying to get mksh through the OSI first now.
kettenis@ said he'll start with the paperwork. thor, our dedicated, is a new
mirror now (also SSL-less,
but I won't advocate not to use TLS).
02.11.2004 by tg@
At work, I tried to build a snapshot, so that our
friend <Kabelaffe> can install that on his fileserver. In the progress,
I synched the lists etc., and it now only waits to be committed, as this wlog
entry. Later, I had an adventure holiday ticket with Deutsche Bahn, by just
trying to travel back to Bonn — but I hacked up enough so that imake
finds cc – or ${CC}, rather. Now I'm sitting at cnuke@'s, writing a
wlog entry and – trying to show him php5's dependencies – fixing
some of that beast. Also, while tail comparing mplayer's shlib dependencies,
I found that I don't have a libcaca.so but a libcaca_pic.a – an
unbearable state ;-) which has been changed now (even if the API is reported
to be not stable yet), enabling mplayer to take advance of it and getting
rid of libcaca_pic.a like with all other _pic libs. cnuke@ is now watching
"Tantei Gakuen Q" (Detective Academy Q) anime I brought him.
Make the website nicer, shrink, etc. and install a RAID-5 on Kabelaffe's
server – his wd3 (primary master) is UDMA-6 even, and wd0 wd1 wd2 wd4
(the RAID discs) are UDMA-5. Performance figures will follow.
Looks
like bsiegert@ started merging OpenBSE ports... *sigh* I hope this is the
last time we've got to do a 'full' import.
03.11.2004 by tg@
I'm totally confused by UTC<->CEST<->CET transitions
and went off from work 1 hour early today. Oops. My cow-orker didn't say
anything so I guess it was okay. I discovered that OpenBSD has started to
use "x.org" as well which has no reason to exist at all IMHO.
Well, synch'd their latest sane code, that's it. I also added some diffs
from tech@openbsd.org which show a disc I/O speedup on odem, in the hope
they don't break, and looked at OpenBSD's binutils-2.15 merger shortly (tho
it doesn't help, since we're using binutils-current). Then I merged your
daily bug of fixes and changes (passwd, ksh, pppd) and fixed patch (also from
OpenBSD). I pulled a diff from gcc-current which renames the java™
related tools and started working on php5 (and its dependencies) to satisfy
the MirCustomers HaMa (Benedikt Steinbusch) and zirk (Benni Adams). Maybe
others will benefit from that as well - bsiegert@ already needs not fix all
the dependencies of php4 but must read my diffs – having different
paths between MirOS httpd and OpenBSD Apache™ is a PITA. I also hacked
support into indent(1) to not touch comments at all which is kinda
cool when converting old code to KNF (someone reads mksh here?). Finally, I
did a port of F-Prot (the personal-use
freeware BSD version) and wrote some wlog entry, discovering I'm up too long
again. I ought to sleep more – in the morning?
On the political
situation: I don't care too much about USA, except when it could affect us.
And I think that, if Bush won, it might affect European laws and directives.
So let's hope he gets a few cluons and doesn't abuse his might again. Has he
been elected at all? I don't know, only what cow-orkers said today; I don't
care about mass media, so... Maybe this time, he's got the voter's majority
though. For me, fighting EU patent and IP laws is more important.
05.11.2004 by tg@
Sort of a small update. I've been in Netzladen.org
again and held a spontaneous presentation/talk about the editors ed(1) and – my favourite – jupp, mentioning that I'm faster
using ed(1) than vi(1), how vi(1) came to be and how vim and emacs failed,
how to remove whitespace at end of line, etc. After the talk, we had great
jugoslawic food (Cevapcici, Pleskavica, Steak etc.) and I drove home by
bicycle late. Too late, since I've got to work today. Yesterday, I hacked
up a dynamically linked c-client, greatly reducing the size of all the
other uw packages, and today HaMa provided us with pear-DB for PHP. I've
done a W3 Validator on our webpages (most, I think) again and they're OK,
now we only need a damn structure for these. We have too many categories
and mis-leading stuff.
06.11.2004 by tg@
Ok, admittedly written on 07.11. early... I'm
playing right now a bit with ancient Wine of 1999, after having written a
server implementation for the heartbeat protocol, hacked up the client a
bit and installed MirDNS #2 on our servers. Oh, a propos: we're
currently off DNS because our parent servers have accidentally removed the
glue to our subdomain. I've also shrunk the MirPorts makefiles a bit.
07.11.2004 by tg@
I've gotten around to fix all issues regarding the
heartbeat implementation and MirDNS – except for us being off DNS at
the moment. I hope ScottyTM will fix that tomorrow. And someone come clean
up my kitchen ;-) ducks and runs
Update: hunting
the remaining bugs isn't fun when my machines (thor and loki) crash one on
one, but DNS seems to be back. Thanks to Andreas Kuper for error recovery.
07.11.2004 by bsiegert@
Begin merging OpenBSD ports in a private Subversion repository. This shall
hopefully reduce "transition times" when MirPorts is broken after an import.
Some small changes at the pkgtools, and a new [link] list of ideas
which should be
implemented. The usual small port updates and fixes.
I was eating out with a friend today at the Flam's. They have great
Alsatian Flammekueche, but
unfortunately I have eaten too many. For those that don't know, Flammekueche
are a vaguely pizza-like Alsatian speciality, made of bread dough,
crème fraîche, sour cream, onions and bacon.
12.11.2004 by tg@
Play with Wine (the old 1999 version); doesn't work
as I'd like it to though. Clean up nc(1) and the hbSuite, adding support for
the "sender" function and basically rewriting heartbeat. Remove
mail check functionality from ksh, clean up and release mirbsdksh 1.16;
jupp 3.1j2 and another mirmake experimental snapshot. Jupp is now maintained
in-tree as contrib too. Fixes for mkinstalldirs under GNU bash. Restructure
this damn website. Try to hack gcc 3.4; give up. Drink too much.
early 13.11.2004 by tg@
Giving up on gcc 3.4 obviously influences the
future developmant plan of MirOS. The new plan is: I'll first bring out the
"last stable snapshot" (containing OpenBSD 3.6 binary emul), then
hack & port cvs 1.12, slow and soft migrate to newcvs (updating binutils,
tagging openbsd during that), except for src/gnu/gcc. Then I'll contribify
gcc 3.4.4 (or so) and offer that as additional package. MirOS thus is not
fully self-hosted any longer, but src/gnu/ isn't considered crucial to MirOS
anyway (and will leave base8.ngz for gnu8.ngz, joining gcc8.ngz which is even
less free). The other alternatives would have been to leave a hacked gcc in
the tree (ugly) or even switch to TenDRA (I don't have the time to maintain
machine descriptions for two different compilers), providing gcc as port;
I'm still open for the latter, just not until after MirOS #10.
late 13.11.2004 by tg@
New day, new fun. I seem to not be able to get
rid of my headaches, though. I've resurrected sparc and mirppc from the dead
to the undead, lest to be seen what we do with it. Also, I've brought the
openbsdulator up to -current, at least I hope so, and went through all of the
OpenBSD flag days and major bumps they recently had. To be verified. After
that, I'll build the "last known good" snapshot, before starting ncvs 1.12
porting, and generation of /Cvs on odem.
14.11.2004 by tg@
I'm fixing all the stuff and now going through the
late flag day myself. If everything goes well, I'll publish a (the "last
known good") snapshot. compat_openbsd(8) should be en par with -current
now.
A nice way to express this paragraph is: (p "14.11.2004 by tg@"
(br) "I'm fixing all the stuff and now going through the late flag day
myself. If everything goes well, I'll publish a (the "last known
good") snapshot. compat_openbsd(8) should be en par with -current
now." (br) "A nice way..." (br) "Nice, eh? :-)")
Nice, eh? :-)
15.11.2004 by tg@
I watched too much anime yesterday - to be exact,
anime with real human beings ("live-action" is what they seem to
call it), namely Gokusen. That's why I slept only 4 hours this night, but
strangely I was pretty awake the next day (today); couldn't sleep in the
tram to the train station, and didn't sleep in the train either, chatting
with an acquaintance of mine. Then, at work, my "boss" was back,
and I had to fuck up Apache™, PHP and Empress DB again, to no success
(well, the CLI works, I'm now going to use the CGI in httpd instead of doing
a module). Except for my ability to concentrate and eyesight both decreasing
after lunch, I was still crazily awake, and didn't sleep too well on the
train back, read a John Grisham book. During all this, odem built a snapshot
of -current, but I'm too lazy to upload it now; I'll probably get rid of the
libgcc.so first altogether, and build the new snapshot in a chroot, as if it
were a release. Tonight, I also read the new Chaosic Rune and
Hunter x Hunter chapters, watched Naruto 108 (this time by [DB] not
[ANBU] for fun), peeked into (and discarded) Full Moon wo Sagashite, and got
to watch the Meitantei Conan movie 6 finally. Whew. It's 22:28, and I've got
to get up tomorrow at 05:00 UTC... can't be helped.
19.11.2004 by tg@
Whoa, 4 days already passed. I've been being eaten
up by making Apache™+PHP+Empress on SuSE GNU/Linux 6.4 work at work,
and haven't gotten around to much else due to lack of sleep. Shadow Skill is
cool, and the recently started Manga-translation GASH! from Egmont Verlag is
worth a laugh (though Conan #32 is probably more expected). I've build a
snapshot myself, installed it on wbx@s machine and tested it a little bit,
and have now fixed all identifyable problems with the snapshot. This means,
I'll build a public full snapshot tomorrow, after I got up.
late 20.11.2004 by tg@
I've started a page featuring a detailed list of changes between snapshots, to aid
bsiegert@ in creating the short list of changes for Freshmeat. I've fixed the
damn 8bit support for sendmail, and am compiling said snapshot right now;
will fix patch(1) in a jiffy and start hacking cvs RSN.
21.11.2004 by tg@
Update patch(1) to OpenBSD version. Fix mis-use of
touch on ramdisks, and mv in X-Window/Makefile. Start hacking on CVS 1.12:
clean up, replace GFDL'd file by free one; shrink. One
caveat: merely trying to access a CVS-1.12 repo
with a CVS-1.11 (or early 1.12) client will cause data loss or
corruption. While hacking CVS, I discovered a grave typo in the
time handling (<time.h> contains macros for tai2timet and timet2tai
which were formerly functions) and revoked the snapshot; will redo.
Ports: mplayer doesn't do runtime CPU detection any more, should improve
stability. metaauto handles autom4te and automake now; automake 1.9 gets
ported. Update some ports (rsync, win32-codecs, sqlite3) and djbdns' root
zone cache data; fix others (broken due to infrastructural changes etc).
I went almost crazy, staying up until way past 5 o'clock UTC, seeking to fix
these damned time management bugs. We're passing the FSF test suite now,
almost (since it's internally tai64_t, we have got only 62 bit + sign, not
63 bit, but it looks good now). Bumping libc major and recompiling.
23.11.2004 by tg@
I'm so tired from sleeping too few I can barely hold
my eyes open, so hold your breath a bit longer for the new snapshot. There is
now, finally, an IPv6 subnet with decent DNS on thor, VPNs next. I've fixed
well enough so that the FSF mktime() test suite passes with 62 bits + sign,
and almost passes with the full int64_t time scale. Also, I've fixed the
fallout I could easily see and synched a bit against OpenBSD. The snapshot
can now be built.
Yesterday, I've been in the Netzladen again, bashing
the mach-ackers and uploaded
some binaries of
my software for Mac OSX, in case your target iMac (iih, really) has got
no gcc installed. I can't imagine anybody surviving without mksh, ed and jupp
around... (bsiegert@ will probably bash me for this wlog entry, but OTOH he
installed irssi (yuck!) on thor, so I can take my time to bash, too!)
24.11.2004 by tg@
Originally, I hadn't planned to use my laptop today,
doing housework instead. Apparently, doing housework is not my favourite thing
to do, and I'm not in the mood for it either. So I went to have a look at my
INBOX, discuss open source licences and IRC. Wait! eMail INBOX? Thought so.
Actually, pine refused to validate the SSL certificate of the servers (both
IMAP on loki and SMTP on odem/thor). WTF? Why?
Turns out openssl is not
only weirdly licenced, but crap too. For now, I've fixed the bugs, but let me
repeat that it is important that all of you DO test
the new snapshot once it is out, and report back bugs. I've also written a
quick howto-fix-two-frequent-time_t-problems in a commit message in MirPorts
today – this is actually pretty crucial since not only MirPorts, but
also MirOS itself, will run on ILP32 as well as LP64, little endian as well
as big endian (no PDP endian tho), 32-bit as well as 64-bit, time_t-long as
well as time_t-64 systems.
Two things I ask for: some girl interested
in a computer geek, and someone interested in funding my travel expenses to
Berlin (21C3), Luxembourg-Kirchberg (linuxdays.lu), Bruxelles (FOSDEM 2005)
and Karlsruhe (LinuxTag 2005)? It's not too much, about 50 € per travel.
People interested mail me or try to use Paypal (which I haven't used yet myself, though, so please
drop me an eMail if you've donated, so I
can check that (first few donations only).
26.11.2004 by tg@
I actually have drunk a whole bottle of wine without
feeling drunk yesterday. Weird, normally I get drunk really well after really
few wine. Anyway, we had a nice Netzladen evening, except that me playing
with OpenVPN was interrupted by the unreachability of my server. I've fixed
OpenVPN to detect dl functions, more libssl format strings, and added users
for ftpd-privsep and openvpn. It didn't work though. I also synched GENERIC
with OpenBSD (disabling kernfs and the layer filesystems), then updated quite
a few things from OpenBSD (ksh, dhclient, isakmpd, ssh, ntpd) and fixed just
another field-separating bug in mksh. I even added a regression test for the
manpage documented bug fixed by Jared in OpenBSD. I'm continuing to clear
MirOS legally as well; the discussion between developers is private though.
I'll take odem with me this evening, containing about 9 GB of Anime and a
copy of /ocvs from loki, and build the long-awaited snapshot (or at least
I hope to do so). This means, though, that either I can't commit there (not
good, since at least the distrib sets aren't synched), or I'll take the CVS
master property with me (which means nobody else can commit until I return
tomorrow at noon or so, but that's not too evil since the others aren't even
writing wlog entries ;-) Maybe they wouldn't even notice...
Setting the
date to year 2050 or so still doesn't work; worse, the kernel doesn't even
boot when the BIOS is set to 2050 (which works perfectly with MS-DOS).
28.11.2004 by tg@
I've had an internet-less weekend at my grandmother's
place, with great food cooked by herself (with a bit of assistance by me) and
ripped out quite a few bugs, building a snapshot for myself. OTOH I'm not too
content with the code yet...
Indeed. I found that some files which are
generated still were left in the tree, and plugged a <bsd.man.mk> hole.
Also, I removed some obsolete stuff and changed the mode of /var/log/messages
to root:wheel 0640. I've also cleaned up src/gnu/ largely, with -Werror all
over most places, and a better <bsd.cfwrap.mk> relying on a wrapper to
the C compiler. Finally, diff and RAIDframe were synched to bugfix.
30.11.2004 by tg@
Today's motto is: Gimme a break.
I've finally
thrown out the snapshot after a bit of
help by the tracker admin, and a night full of builds (and curses about a
missing CC=$CC in a Makefile). There was a timecounter sneaked in in the
RAIDframe update, and some more time_t vs long vs int (really!) bugs in pine
which kept me busy (I'm just mentioning this here because tyler@ already bugs
me enough with his cry for gcc/g++)... also, I'm kind of the only guy doing
the website, and playing a bit on license-discuss@opensource.org lately.
After the snapshot I cleaned up a bit, removed <bsd.regress.mk> and
switched NOMAN, NOPROG and friends from defined/UNDEF to YES/no. (Oh, NOPROG
is actually not used in the code.) This yielded a speedup in make(1)s, and
pointed me to the libedit readline emulation, so I got rid of GNU libreadline
tonight. Also, I've fixed some stuff I thought I had committed earlier to
ports/Setup.sh and enforced the existence of an (empty) libdl, just like with
libresolv. This helps GNU autotools to find dlopen() in libdl (silly).
01.12.2004 by tg@
Just when you think it works, and install a recent
OpenBSD 3.6-current snapshot into chroot /emul/openbsd... you discover that
struct rusage contains timevals. And I've added some birthdays and events
to calendar(1), and fixed ports using schily.
02.12.2004 by tg@ – Part I
Trying to commit this via ncvs and
:ext: to loki running ocvs. I've hacked the time stuff I found yesterday
(some more missing syscalls for compat_time_) and switched my box to cvs
1.12 (as ncvs, not yet cvs). I should finally fix herc, write some more docs
and licencing info. First -> netzladen.
02.12.2004 by tg@ – Part II
Add a great deal of emul pathname
translation to the Openbsuxulator and bring a certain trademarked webbloat,
erm webbrowser to work. The rest of the evening, try to make fonts look
good and curse fontconfig again.
03.12.2004 by tg@
Heavy porting effort leads to successful integration
of cvs 1.12 into the system and already akomplished patches to the log_accum2,
commit_prep2 and (new) tag_accum2 scripts. I ought to sleep more.
07.12.2004 by tg@
There have been many birthdays apparently. Geier of
BOSng, Halil (a former classmate of cnuke@), yofuh of BOSng, bsiegert@ today
and in a few days my little brother reaches the 20. Anyway, back to work,
except that I need a break (again). Herc is still not up, but I'm pretty lucky
with GNU CVS 1.12 – except that there appeared OpenCVS (by OpenBSD) in yesterday's CVS
commit mails the first time publicly. This sucks. And no, I'm not going to
take over the OpenBSD way, since I had to update the Netzladen-Router to
kernel PPPoE yesterday which is so unusable I can only recommend to
not use it if possible. I also thought of mksh-ifying log_accum2 and
commit_prep2.
cvs 1.11 in #7-stable and cvs 1.12 are now quite
interoperable: they behave almost the same and seem to be able to talk to
each other via the network sanely; they won't access each other's repo
unless read-only. -R is still not passed over the network, though.
Both ifstated(8) and ifwatchd(8) are huge failures; the latter just doesn't
get any messages from the routing socket for UP/DOWN, just create/destroy.
Looks like it's OpenBSD's fauly.
I removed traces of two old eMail
accounts I used and updated some licences on the way, and also polished
<sys.mk> a bit. compat_openbsd(8) now uses kern.emul_uname sysctl,
a full sysctl number translation is still missing. mksh v1.17 is out. I've nuked
csh, added a GNU libreadline port (the first to banish the GFDL'd docs
from being redistributed violating the licence, and adding _gplwarn.c
to the port being built) and had python pull it. diff, libz and libXpm
got updates. Useless stuff was removed, and the original K&R added.
We build HTML versions of the papers now as well. Benny uses the Plan9
shell "rc" for a while now and updated the port.
I'm going
to pay our "FH Bonn/Rhein-Sieg" (approx. a university of
applied sciences, according to sap@) a visit tomorrow, got even free of
workschool, to look what I'll be (probably) taking after apprenticeship.
What crazy GNU freaks and Mac hackers there are, and how evil all the
bureaucracy and professors are, is probably what they won't show me yet,
but I think I can compare it to the IHK training I get now. I won't like
it, but I'll try to have friends, have fun, meet a girl (?) and take with
me as much knowledge as possible.
Did I say already OpenBSD sucks?
Their routing socket is incomplete, Kernel PPPoE unusable, the ($ext_if)
syntax in pf.conf(5) doesn't work (in MirOS) and to publish OpenCVS right
now an affront (they must have known I was working on GNU CVS,
which we'll stay with).
And looks I'm still the only one doing commits
here, especially in www/ — at least we'll soon have another redirector
for our websites, a permanent cvs[web].mirbsd.de (I hope).
Now I'm
waiting for somebody to donate some beer, wine, Met or equipment...
Update: I removed some features from pf now which I always rant about.
07.12.2004 by sap@
Not much news, but tg@ wants some company in
the wlog. I'll apologize in advance for any html mistakes on my part.
My schedule has gotten a bit quieter lately, though my programming time is
limited by my one year old daughter's fascination with the computer (and
natural desire to bang on the keyboard as I type). I'm rebuilding MirOS
on my workstation for the first time in a couple of months, then I might
do some MTA related work (also for the first time in a couple of months).
Time to go to bed now though, have to be rested for the dayjob.
08.12.2004 by tg@
Today I paid the
local University of Applied Sciences (approx.) a visit, learning about
how you study there, visiting some of the readings/presentations (pretty
bad compared to what you learn during my apprenticeship, but overall on a
much higher technical niveau) and talked to students (they use the verb
"Studierende" instead of "Studenten" which sounds to me like they are
"studying people" which cannot be because "lazy student" is a tautology).
Looks like I'll kill my time there for the next three years, 2005-2008.
Maybe I'll take a Master (magister) after the first exam, the
Bachelor (bakkalaureus) too; then I'm probably 30+ when I finished.
But to get a nice diploma-like thing and more money later might pay off,
and learning about red-black trees, IEEE floating point numbers and finally
being able to design a malloc() algorithm should be nice for a BSD developer
as well. Benny, Stephen, Tyler, I'll join you. If I can pay it.
09.12.2004 by tg@
Today, I was like I need vacaciones. I'm lustless
and couldn't get around to do anything. That's why I went destructive and
first unsubscribed some usenet newsgroups and IRC channels, then made the
entire /usr/src/games/ a port (ports/games/bsdgames) and nuked it and a lot
of other stuff. Then I started rebuilding, to look if anything broke. Some
OpenBSD updates went in, and mksh
R18 was released.
10.12.2004 by tg@
Backport some security (UFS, FFS) and bug (KSH)
fixes from OpenBSD. Fix the bsdgames port. Fix mksh build on Interix. Fix
$IFS mistake in MirLibtool. Say hello to miod@openbsd.org on our mailing list
;-) Fix -Werror in conf.sh and time_t bugs in GNU RCS (don't know why they
weren't found with -Wformat, might be more). Start :ext:tg@odem:/cvs. Bring
out mksh R19. Reduce diff with OpenBSD in base (/ocvs). Nuke remnants of
profiling code, deprecate getopt(1) in favour of ksh(1) getopts builtin. Add
group 'audio'. Clean up distrib lists for ramdisks. Make pine more usable and
standards-conformant: default to downgrade-multipart-to-text.
11.12.2004 by tg@
Finish code-removal actions; build a snapshot for me
internally. Now the one thing missing before the ncvs move is to downgrade
MirOS httpd from ASF httpd 1.3.33 to ASF httpd 1.3.29 and upgrade it with the
recent obsd changes. Wish me luck (especially to open that wine bottle).
12.12.2004 by tg@
After (yesterday, late) and before killing some wine
I've done a big-bang 亀はめハ (kame-hame-ha)^Wcommit,
updating MirOS httpd with OpenBSD, downgrading to ASF 1.3.29 first for
licence reasons; this also will shrink our /ncvs later. I've also done some
necessary steps for ncvs: removed the cvs->ocvs compat symlinks and moved
the web URIs of the ChangeLog files. Some fixes in the helpers. Someone sent
us fixes for ifwatchd(8) with sppp/pppoe in-kernel. Other than that, I just
found out that, with our new snapshot, we save about 4 MB compressed (not so
much if taking into account that the papers are now in htman, too).
14.12.2004 by sap@
Nothing too exciting happening. Poked around
in ksh (testing the waters, my interactive shell is usually tcsh), ended
up installing mksh R20 on my Mac and updating the wiki page a bit.
Enjoying unseasonably warm weather (5 degrees C above 0 :) and making
good progress at the dayjob, if not at spare time programming.
17.12.2004 by bsiegert@
Yes, I haven't written any entries here for a long time. But I had an Organic
Chemistry lab course which was keeping me incredibly busy. Now, the christmas
holidays have begun for me. My desktop machine has a new PSU now, so it
should work flawlessly.
I have continued work on the pkgtools, adding @man support and fixing
@lib (and find_plist()) for good. make update-plist
now correctly inserts @lib statements. Thus, when updating a port,
remember to update the PLIST. The @lib thing will be really important
once we get package updates. What remains as a showstopper now is the
dir/ implementation which needs to be redone with support for
@sample dir/ and owner / mode changes. Oh, and I have
begun to write a bit of Objective-C for what will become the new PLIST
handling. But there is still nothing to show.
-8.01.2005 by tg@
No, not kidding. My laptop is currently showing me this date. I'm sitting
at the DECstation 5000/260 (NetBSD™/pmax 1.6.1, mirbsdksh) in the
Netzladen and not capable yet to write a wlog catchup entry since I'm
overworked, but I thought I'd share some updates.
I've actually fixed the date problem, but cannot rebuild src due to
lack of an actual working /usr/src of -current (mostly because parts of
-current are in ocvs, parts in ncvs and parts not in any CVS but in my
merge checkout working directory tree). I've replaced parts of gmtime(3)
and localtime(3) with the DJB
functions. I think I'll just do the similar for mktime(3)'s side effect
struct tm adjustment function, but this needs research.
I now know that I don't get lent a car for driving to 21C3, but I at
least know where to sleep, and Zeno (Przemysław aka Przemek)
just wants a Club-Mate and a Kölsch for that. Now I've get to beg
enough people to take me with them, because I cannot afford Deutsche
Bahn (I can't afford the whole congress!).
We'll be burning and selling CDs of MirOS again, but this time nice boy
Angelo of MachAckers Bonn
offers to burn them in his G5. The content will probably be a dual
setup CD with the 20041129 snapshot and some #7-stable Benny is going
to prepare on it (if not, I will take #7quater). Maybe #7quater/sparc
if that one fits.
Yesterday was the first day of the new year, and it's in the sign of
having avoided the dreadful European software patent directive. Well,
more like slowed down, but in this spirit, watching my candles, I
greet all of you readers and wish you a happy new year (according to
the winter solstice calendar), and to have a nice jule fest (for the
Asatrú) or Christmas (for probably most of the readers), or
light celebration / family days, for these who join the christian
majority in their countries.
04.01.2005 by bsiegert@ (in the train to Strasbourg)
So the holidays and 21C3
are over, and it is just three weeks until the exams *sigh*. At least,
Christmas got me a new external 60GiB hard drive :). I also got around to
hacking more on Objective-C in the train from and to Berlin. And let's not
forget the web site with the photos from the
#deutsch channel meeting.
The two projects for 21C3 were: MirOS BSD #7quinquies and the macppc port.
What has become of them?
During the congress, I remotely built and installed #7-stable on amiras,
which was happily sitting in France. However, in the last stage just before
the final upload, the system seems to have locked up. Thus, there is
something in the v7.5 directory on the mirrors, but it is
not yet the final #7quinquies. The binaries will be rerolled
in a few days, so don't download this yet. Wait for the release announcement
on freshmeat and on the
miros-discuss@ mailing list.
MirBSD/macppc aka MirPPC is a mixed success now. GENERIC panics on boot.
However, there is a SMALL kernel with some features (mostly ram disk hooks)
disabled that is bootable. Thorsten has cross-built a userland using the
wonderful MirOS cross-building infrastructure. However, it was compiled
for the PPC 7450 (aka G4), so I get lots of Illegal instruction
errors on my 750FX (G3). But I will try again with a new userland to
manage to build a snapshot natively. This one will be all static without
support for shared libraries which will come at a later date.
CVS is a mess right now. All PowerPC development takes place in the
MIROS_X branch of ocvs. #7-stable resides in the MIRBSD_7
branch. thor is ocvs master. Some parts of the system have already been
moved to ncvs.
Don't know which date I should write here, I've been wake since 50 hours.
05.01.2004 by tg@
So much for the wait for a catch-up wlog entry.
I guess I'll deliver one when I get the time.
I've built gcc 3.4.4 (yes,
I know that it does not exist yet, but this is what it tells me about its own
version number) with all languages and propolice, boehm-gc (passes tests),
libffi, libada, libobjc on i386 so far. There have been not even problems
with us defaulting to -fstack-protector-all (hi etoh-sensei in case
you read this). I've disabled ProPolice SSP for the entirety of GNAT because
Ada seems to have its own stack smashing protector and that baby is
crashing with an ICE when ProPolice is enabled. (Maybe etoh-sensei develops
on OpenBSD, where Espie hasn't delivered a gnat1 since their switch to ELF.)
I've also discovered a solution which renders the whole idea of using
-fno-zero-initialized-in-bss (even if only for the kernel) absurd
(emacs still needs it, hi sap@ in case you read this): just force the needed
variables into .data. I haven't built anything besides gcc itself
and its libraries, the boehm-gc test and a C and Objective-C hello world
(and a C++ hello world with libc not STL) yet, though. I should cross-build
for mirppc (and write mirsparc machine descriptions), but I guess I'd rather
move over the rest of the userland, so MirOS X is gonna be developed on ncvs
not ocvs (although I suppose this will slow down bsiegert@, sorry lad).
I've nearly recovered from the flu I took home from Berlin (21C3) and hope to
be able to go swimming on friday. Next monday, I've got to work for a living
again. (Oh, in case someone'd employ me between mid-June and end of August,
just tell me.) The MirShirts #2 have been not quite the success I had planned,
so I'm considering to sell them (additionally) via Geekladen now. That will cost more
(and ask more effort from me) though, but maybe pupils still can get the 1
€ discount. お休み。
07.01.2005 by tg@
A more regular entry. I've merged X-Window so it gets
out of my feet and I more free disc space, as well as all the old stuff from
NetBSD™ and stuff without a vendor branch. Looks like this wlog section
can be laid ad acta after about 200 commits or so soon. I've also packed a CGZ
for these who'd like to peek into ncvs really quick (mksh doesn't build on
non-MirOS atm though, and the security fixes and merges from obsd are on my
TODO, admittedly). The kernel builds on gcc 3.4 and I have no idea why OpenBSD
needs to circumvent the default -fzero-initialized-in-bss (it only affects
static variables for me after all). Our kern_timeout.c is now derived
from NetBSD™-current, even. I guess this means (for Éric
Lévénez) to draw from NetBSD™ 2.0 (he doesn't have any
HEAD branches) and OpenBSD 3.5 to MirOS #8 (not yet in existance), since we
take only the most important fixes and some apps (such as OpenNTPD) from the
newer OpenBSD sources. I'll go swimming tomorrow, and work starts next week,
so expect the commit rate to slow down a bit ;-) Luckily, the automatic
history and ChangeLog file rotation now works (tested).
08.01.2005 by tg@
Today, I've written a rather large status report
which sums up quite well the current state of the tree. I figured this would
be easier than to write up all the changes since the last wlog entry of 2004
;-)
Swimming was fun, but everything aches now...
14.01.2005 by bsiegert@
My exams are coming nearer and nearer. In the week from January 24, I will
have nine (!): crystallography, fluid mechanics, molecular symmetry, materials
science, statistical thermodynamics, statistics, organic and inorganic
chemistry. This means that for the rest of January, I won't be able to do much
MirOS-related work.
There were some security updates in MirPorts: CUPS, Icecast, and zip. While
updating the PLIST for cups, I discovered a bug in the package tools: Before,
they just inserted the ldconfig command at the end. If there was a
@cwd directive before (a lot of ports do this to change into
${SYSCONFDIR}), the wrong directory (/etc/lib in this case)
would be used for ldconfig. This is fixed now, so use make
setup before installing the updates.
15.01.2005 by tg@ – Read this if you come
here from "bsdforen.de"
Ich finde es immer wieder lustig, wie
manche Leute fundiertes Halbwissen verbreiten, aber die neuesten
Errungenschaften des bekannten deutschen Deppenforums sind nicht Halbwissen
sondern Unwissen. Ich finde es zwar lobenswert von templis, über uns zu
schreiben, aber das strotzt nur so von "IIRC" und ist von vorne bis hinten
falsch.
MirOS ist ein Projekt, das aus dem System MirBSD (src+ports)
entstanden ist und von zur Zeit sieben Kernmitgliedern entwickelt wird. Das
MirOS-Projekt liefert nicht nur MirOS BSD und MirPorts aus, sondern auch
die MirbsdKsh und diverse weitere Progrämmchen. MirOS ist das bessere
OpenBSD und wird auch so entwickelt – da ich von OpenBSD-Seite her
hauptsächlich ignoriert werde, sehe ich nicht mehr ein, Patches selber
herauszureißen und in gegen OpenBSD integrierter Form an Nutzer von
OpenBSD zu schicken, denn hier schneide ich mich ins eigene Fleisch. Diese
Leute sollten entweder MirOS nutzen oder (zum Beispiel wenn sie Features
brauchen, die MirOS nicht bietet, wie Kerberos oder NIS/YP) selber porten.
Oder mich dafür bezahlen, daß ich das mache (ich stelle auch
Rechnungen aus).
Ein Openbsuxulator existiert zwar, ist aber
bloß eine Tabelle, in der den OpenBSD-Syscalls die entsprechenden
Wrapper um die MirOS-Kernelfunktionen zugewiesen werden (hauptsächlich
weil wir bereits mit 64 Bit Uhrzeiten arbeiten) und funktioniert genau so
wie der bekannte Linuxulator, nur
besser und gegen OpenBSD 3.6-current feature-komplett (bis auf LFS und XFS).
Ich nutze zum Beispiel centericq in diesem.
SMP ist keine kritische
Neuerung, sondern schlichtweg Unfug – ich werde mich hier nicht weiter
zur OpenBSD-Implementation davon äußern (zumal ich weiß,
woher das Geld dafür kam, und das gut von den dunkelrosa Jungs finde),
hingegen sind enorm viele kritische Änderungen vorgenommen worden, wie
der 64-bittige time_t, gcc 3.4, die CSU von NetBSD™, auch das Schaufeln
von Bibliotheken (png, expat, cdk, readline), neue Binutils, neues CVS sind
nur die jüngsten Beispiele. Wer weiterdiskutieren möchte, bediene
sich doch bitte eines geeigneteren Mediums, denn das WWW ist für
so etwas ungeeignet.
Nachtrag: ISDN mit OpenBSD 2.8 hat nie korrekt
funktioniert und kann auch mit der Fritz!Card nicht umgehen. Hellmuth hat
wegen mangelnden Interesses von seiten der OpenBSDler aufgegeben.
15.01.2005 by tg@
I've now finally managed to set up OpenVPN between
odem and thor, but there seem to be kernel bugs regarding IPv6 address setup
and removal (and there are some OpenVPN bugs, too). Stephen has discovered a
shortcoming in CVS so severe that we have got to re-do ncvs, as already shown
on the mailing list. The remaining merge is progressing slowly but with great
care, and you can expect a real surprist for ncvs ][ too.
17.01.2005 by tg@
I guess it's time for another wlog entry. Today,
which started on January 16th (no I didn't even sleep half an hour in the
meantime) I've kicked my lazy bum and started to port binutils – it
was mostly the usual dealing-with-autotools stuff and some -Werror fixes.
I've also thrown a bunch of other bug fixes into binutils and gcc and set
further up my OpenVPN (now the client runs as user _openvpn chroot'd into
/var/empty as well, and the route and IP address handling is totally done
by mksh scripts, powered by the new /usr/bin/route_print tool). Because I
had to reboot quite often (may be kernel bugs, see the mail to the list),
glad to have softupdates, thank you, I eventually discovered that –
no matter if the kernel was compiled with gcc 3.2.3 or 3.4.4 – ntpd
started with a 50 second offset which led to a mathematical algorithm as
well as empirical code redesign for some stuff, discovering some apparent
shortcomings in gcc which I couldn't even track down, how bad. Today I've
finished this kernel work – it seems like the time is correct after
a boot-up again now. As a side effect the code is more clean, I nuked two
files and the kernel builds with mgcc – oh wait! what about "I
can't pack your struct" attitude of gcc? For now, I'd just ifdef out
the __packed for _LP64.
Okay then, back to my binutils: bfd, opcodes
and from gas back to bfd, playing with autotools, ld and finally binutils
themselfes. I've left out gdb, it's too major effort, and the new patches
weren't applied either yet. During that, I've also thrown in a good bunch
of OpenBSD fixes which I already had stacked in various mailboxes. After
that all, I started a make build (not a release, it's still intentionally
broken) and walked that through, finally discovering the #pragma pack(1).
D'oh! And it seems to work.
All in all, gcc 3.4 leads to better, yet
more inefficient code, and I've even played with the COPTS again to fight
the code growth. I've been doing 3 to 5 things concurrently since after I
got back from the overnight visit at yofuh's place (where the OpenVPN and
time incidents took place), and still feeling great until about midnight,
UTC (kay, that's probably caused by the second of my beer bottles nearing
emptiness – I don't drink much, despite what people say and despite
our OpenBSD origin). This means we're about to enter the #8-prebeta stage
in development, with the release process – in particular the kernel
configuration and build system – still being the redesign candidate
number one. In #8-beta, there's updating gcc, binutils and lynx (and some
other stuff if required), importing from OpenBSD, fixing and writing docs
for real – and most importantly testing the new gcc for ports. Then
there's macppc and sparc for #8-gamma, during which the MI and i386 parts
should be already quite stabilised, and all thumbs up for #8 awaiting the
MirPorts peoples' readiness. (I'd like a full package build, with all the
broken packages marked as such.)
If you look at the HTML source code
for this wlog entry, it's written as justified paragraph with no cheating
(double spaces) applied, and in fact it's become quite natural to me. Two
times I had to stop, think and re-verb my sentence, four times words were
exchanged to fit better, but overall it's been written quite fluently. As
I'm writing, there's another bug in zic which goes away if I insert a few
printf for debugging. Weird stuff, that.
19.01.2005 by tg@
I've finally been able to do a make build on src
and make b-r on gcc. src doesn't succeed a release yet because crtbegin.o
and its cousins are (usually) part of gcc. The solution is clear, write a
b-r shell script which does src:build, gcc:build, src:release with crt*.o
copied from gcc, gcc:release, htman, X11. Or (probably better) src:build,
gcc:b-r, src:release, htman, X11. I've also found Reiser CPP (from 4.4BSD
too, like nroff and friends) which we will use (in X11 for example) until
the gcc module's GNU cpp is installed. Another lot of minor bugs I fixed,
and there might be lurking more – watch out for everything that has
its own subdirs, such as usr.sbin/ppp :-(
Today, I've got my project
appliance ready to submit so far and received my personnel review at $ork
which was surprisingly good. I've had some stuff to take care of this eve
so the new b-r script I planned for tonight isn't ready yet and right now
I'm too tired, so I rather went for Anime and RSN bed. I'm not doing much
for ports right now (I haven't even run 'make setup' for ages) tho.
22.01.2005 by tg@
Thundersday went quite well, so I had a fruitful
and chilly evening at the Netzladen. I was going to up to gcc-20050121 on
Friday, but I slept not well nor much hosting wbx@ for a night, had to do
over-time at work and was absolutely down physically and even ill late. I
then slept too much until today and got up too late, so I was again quite
sick. We had a meeting at the Netzladen of one of my usenet groups (which
bsiegert@ also idles in), proving that it still lives, and during which I
finally got around to building gcc and then base. Back home, I doped make
b-r for base and started building snapshot tarballs (well, cpio balls but
does cpio have them? :). That'll lock /ncvs and be installed on herc, for
we all await ncvs ][ now – which will start with a real surprise at
the third or so commit, aiming at my fellow developers. Stay tuned.
23.01.2005 by tg@
Some followup on BSDforen.de: it's getting worse
not better, even if some BSD committers and similar people are
hanging around in the channel. Starting with bashing they're attacking me
personally in the end. I don't understand what was unclear in my previous
message: pay me or I won't port ISDN to OpenBSD. The behaviour or whining
(rather) of yours is like demanding of an OpenBSD developer to port SMP to MirOS, and
it's not a bad comparison because in the meanwhile, many things
have changed in MirOS, with regard to OpenBSD. I think BSDforen.de really
is a bunch of immature adults behaving worse than kids and deserving more
a DDoS than a "fuck-off {in,out} on $all_interfaces" – not that I'd
do it though, this is just my opinion, and I got attacked personally more
than is fun, in absentia.
I'm installing the new snapshot on herc at
the moment, re-doing its parititioning scheme only preserving /home (and,
of course, the better of /etc) from the old herc. I'm recompiling the ssh
client with CFLAGS_clientloop.o=-fno-stack-protector because it sucks, to
see it dump core after each connection (ssh with clientloop.c compiled by
gcc 3.2 and everything else including linkage done by 3.4 seems to work).
Let's see how many ports fail on the new system...
27.01.2005 by tg@
Sitting in the Netzladen, I don't have much time
or want to really write a wlog entry. I've assured like 165 points to the
visitors of today's meeting for CAcert.org, so I'll be able to issue even
25 points RSN. And there's a new snapshot,
including ports available for download from the mirrors as well as file swarming downloads. I have
brought this out even if I didn't originally want to commit into "ncvs I"
again, but because -funit-at-a-time looks problematic especially
in combination with propolice I decided to do it.
More and more lads
in my town are gathering for next Fosdem
2005, the upcoming event. Sad we've got no seats, but at least we can
be there and meet tyler@ in persona. wvdputte@openbsd will probably issue
CAcert points too, this drives a few of them as well. I'm not too ok with
not being able to sell T-Shirts, though – but you can still order a
shirt by eMail; there are still two white MirShirts #1 (both in XL) left,
I sold two this week (thanks!).
This week I got these "I don't wanna
continue any more" feelings again, partly because of the war with the one
worse-than-childish stupid forum already mentioned. (On the topic, one of
their users or admins or something contacted me and tried to play it down
(and some other person which I don't mention, whose fault it is that this
page's HTML source doesn't look as nice any more); he eventually asked of
our users to populate the forum and spread pearls among the pigs (that is
my wording, excuse the translation of a German idiomatic expression), ie.
spead knowledge. I won't do it because I despise web fora, it requires to
sign up for an account first, doesn't work in Lynx, and I don't have time
for stuff like that anyway (miros-changes@ users can guess).) Well, back,
my bad mood. I've been eating good at my grandmother yesterday and really
ought to go outside more – but it's snowing and pretty cold (but it
must be colder than -10 degrees to scare me inside the whole day). I also
could go swimming, but not having a car makes me pay for the train (which
is not too much since I can drive around some area for free at the moment
but still a lot for longer-distance journeys). More problematic is that I
don't have a person with whom I could do this since most people around me
are more or less typical hackers, nerds and geeks. They could probably be
inside in the sub-basement three whole days only going out for a pee, not
even eating anything not from the pizza (/me: eek, cheese) service.
While writing this there are more people wanting CAcert assurance so that
now I've given out a total of 205 points, wow. Now, sleep ;)
28.01.2005 by tg@
We've got IPv6 connectivity again. I applied for
a tunnel to thor and throw an UDP tunnel from there to odem right now; no
connectivity for loki and herc yet but they might follow. I'm powered out
still, waiting for the weekend. For the UDP tunnel, I've raised BUFSIZ in
netcat to 4096, works pretty well now except I still get the freezes, may
be routing related.
Well, looks like the snapshot does well.
An
update: I have to advise against using tun(4) in tun or tap mode with any
other protocol than IPv4, because on destroy, active routes are not freed
correctly and cause crashes later.
03.02.2005 by tg@
bsdforen.de strikes again. Looks like they think
they did a great thing, but nobody I know would ever buy
a blowfish as ugly as this
one. There is nothing good coming out of bsdforen.de, still.
Now
to the MirOS progress: nothing ;-)
Well, except for another round of
CFLAGS fixes, and some new/fixed/updated ports, all minor stuff not worth
to be mentioned.
Netzladen now has a music playing daemon, but IMHO,
the mp3 printer is better ;-)
05.02.2005 by tg@
I've felt obliged to start ncvs ][ after all, so
I did so. When having breakfast yesterday, I lost a piece of a tooth, too
bad. It doesn't hurt (now) only disturbs, but having to see the doctor is
not a thing I like – it will also impact me not only physically but
also financially. For FOSDEM, I can borrow IsCs's car (it's 1337 and does
work, advertising for Club-Mate included, but... it's an adventurous car)
so I'll be at least able to arrive. My herc server could need a passively
cooled PSU in Baby-AT form factor.
07.02.2005 by tg@
Carnival update. I've discovered that benz, even
if he doesn't write wlog entries *hint, hint*, still documents changes in
the MirWiki. For me, tomorrow is a working day at my dayjob – maybe
I'm the only one of my class, I think all the others have booked that day
off to enjoy the last day of carnival, before on wednesday the Christians
will start their 40 days of abstinence. I don't mind working on that day,
since I'm born on "Rosenmontag" I hate carnival like the pest, but due to
my $orkplace being Köln, the #1 point of interest for these crazies,
I'm not too enjoying the idea. I hope that, tomorrow, in the train, there
will be at least no smoking (it's forbidden!) teenager groups mocking the
passengers, like last Thundersday.
Well, CVSweb sorta works and _is_
valid XHTML (where I checked, at least), the synchronisation of the three
CVS repos works, I've got a checkout of src to start the merge, so I'd do
it better soon (since I've got a pile of fixes to commit which I can't at
the moment due to lack of a non-locked repository ;). I'd rather like to,
among other things, play with the new gcc and especially native WineX and
Java™ but since I haven't even recompiled XFree86™ it's still
a long way before I can think of starting this. Ports could use some care
too (the pine update isn't that urgent, but mplayer is, and Samba).
At least I've got a spanish guy (evilwm and Crux user) persuaded, in IRC,
to do docs for us.
Oh, well, IRC. I still wonder about the behaviour
of some people. A person with a three-letter nick (I'm not to name anyone
here again) just left the "unfriendly" channel, without telling
the reason. If it's because of my friendliness towards newbies, he should
have left two or three days earlier, so I just don't understand. If not I
have no idea either. Well, there is quite a bunch of people still hanging
around in the channels, so I guess every community builds itself, and who
doesn't fit in earlier or later leaves anyway. I don't want to gather the
huge crowd around myself or the MirOS project, I don't want (us or me) to
be popular. Linux users are a huge mass (or mess?) but definitively don't
have class (or does the Englishman say style?). I hope that, not despite,
but because of our (well, mine ;) I don't speak for the others.) rudeness
towards idiots, our userbase will consist of people which don't lack more
than one or two clues.
It has come to my attention that this wlog is
actually read by people, even if not provoked by links on certain webfora
(I still won't consider it a blog, but who must can retrieve RSS feeds of
the MirOS mailing lists at gmane). These people are hereby greeted.
08.02.2005 by tg@
Today I'm writing a bit about me and MirOS, and why
I'm doing things like this.
First off, I started BASIC at age 8, on an
8088 Schneider EURO PC. Then came English, Assembly and Pascal at age 12,
followed by C at age 17. I shall be 23 in a few jiffies, and I'm not a student
of computer science. In fact, I believe I couldn't even write a (decent)
malloc implementation. If I could, though, I could do it in any language I
know (well, I won't do it in brainfuck(1)
though).
It has been said I'm an amateur. Well, not exactly, but this Theo
quote can only mean me. I do believe so, too, but I'm content with it.
I'm not a CS student or master, worse than Torvalds in that respect, but
I'm kind of a craftsman, maybe even more an artist. MirOS is not an OS
to please the masses, bring in money or hunt bugs down because that is a
design goad. MirOS is simply a piece of art.
But it goes further
than that. When I'm doing certain things for MirBSD™, it's like magic
to me. The licence
is such an example. It was an artwork to do it, but in contrast to the last
193 lines of HTML source in this file, it was also a magickal item to me. It
does good to the source when it touches it. I'm handling forces older than
mankind when writing code, no, crafting a work of authorship under European
Droit d'Auteur law / Urheberrecht, and putting it under the licence is a
magickal act.
Yes, I believe so.
And I also try to stay alive in
the world (doesn't everybody?) beyond my lifetime, through my achievements.
If the remembrance of my person lives on in my friends, wife and children
(which are yet to be haved), things I built are still in use – then I
have deeply engrained myself into the earth's surface. MirOS is a small piece
out of this masterplan of mine. Feel free to help.
09.02.2005 by tg@
Latest (and hopefully last) chapter of a certain
web forum: I'd like to warn everyone from mail contact with "them" (it is
only one special person, but I've said I'm not going to write names here,
again). He has in fact quoted from private eMail exchange with me. I have
him take it down, and he did it, replying over-polite (i.e. mocking), but
that's enough with me. I don't have the time nor is it pleasure for me to
deal with them (or tell them stuff about MirOS which they don't know yet;
I simply have got better things to do).
Quoting from private eMails,
publicly, is simply The bad thing to do, I hope he's learned it.
09.02.2005 by bsiegert@
I have holidays (sort of) this week and the next one, so finally, I can spare
some time for MirOS. There were some updates I have been wanting to do for a
long time, e.g. WindowMaker and dillo-0.8.4. The
[link] pkgtools have had two real
showstoppers fixed: @sample dir/ works now. With this command, you
can create a directory (e.g. in /etc) during installation of a package which
will be deleted on deinstallation if it is empty. This one and the variant
without slash (for copying configuration files) now correctly apply
permissions. Some bugs that have been lurking for a long time have been
fixed—it is safe now to use the -n parameter to
pkg_add. However, I think I introduced a new bug, I will have to
investigate further.
MirOS #7quinquies (the #7-stable branch) has had its first security update,
consisting of a base system tarball and a kernel update. All #7 users are
encouraged to install this. Get all the info at
MirWiki:SevenQuinquiesUpdates.
Some security updates: exim and mailman. There are still some OPSAs unfixed,
but it is getting better. My next big project is bringing GNOME 2.8 to
MirPorts. I am taking a different approach than before: instead of importing,
I manually pull the patches from OpenBSD via cvs rdiff -D 2004-05-28,
merge, fix, build and test locally before committing. The new-style PLISTs from
OpenBSD actually work very well. The only incompatibility I have noticed is
@conflict and @newuser/@newgroup. I won't implement
the latter two, not even as no-ops. The reason is that the ports who use this
tend to also have @unexec userdel or a similar construct at the end.
Thus, it is better to have pkg_create fail with an error, so that you are
reminded to check the PLIST file.
09.02.2005 by tg@
Hey, what's this, I'm reading drexim here ;-) It
did a good job though today delivering an eMail of mine to the GNU
Pascal mailing list and back to me in... let's say, 2 seconds. (Of course
there was my superb sendmail involved, too ;)
Anyway. I've hacked at
MirLibtool again, discovering that "we don't have relink issues, and
we don't relink anyway" (huh?) and fixed some C++ problems. (Rather,
some no-C++ problems.) MirPorts' infrastructure received an update which
also will cause some spurious rm(1) errors to go away. I should really do
the ncvs2 merger now, I wish I could have a week off (and somebody caring
for me, like cooking, making the dishes etc :) I haven't had any holidays
for ages now.
I'll go to sleep now, but my masterplan is as follows:
First, src/ gets merged (pretty quick, but time consuming, since it's not
few files to be updated). Then, I can at least bring in necessary fixes I
have already piled up on my TODO and work on the infrastructure (ie. that
damn MirLibfool). After that, gcc 3.4 will be taken (another CVS snapshot
of course) and imported, then gradually adjusted to match the one in ncvs
1. There's two commits in OpenBSD for that I think we'll like, too. After
that's finished, I can build another (experimental) snapshot, and then it
depends. I think libstdc++, XFree86, gcj/f77 is the order. I would do X11
later if they brought out 4.5.0 now, but I guess, since 4.4 was imported,
we'll stick with it first (I've got to check if it works at all, too). In
the meanwhile I hope Benny updates all our ports (yayayaya I'll do a few,
too), then we'll close ocvs and ncvs 1 down.
I've read NetBSD™
stuff yesterday (and I'm impressed), and I remember Stephen talking about
making the kernel aware of userland threads. Please, do it (I can't do it
myself since I'm just an amateur ;) because it will also increase,
among other programs, wine's chance to work. (Java™ too.)
11.02.2005 by tg@
I'm sitting here at Kabelaffe's (a friend and an
apprenticeship (he's IT-Systemekeltroniker) comrade) continuing what I've
done over the whole night (rendering me pretty useless at work today with
only 2.5 hours of sleep) – updating MirLibtool to 1.5.12 of GNU and
simplifying and enhancing it during that. I seem to have success, even to
bring back (and conditionalising) Fortran 77 and GJC support, so reducing
the diff to GNU Libtool (which increased again due to accent cleanup). It
works on a few ports, but I need testers now.
I'm still open for the
weekend, having planned only to not plan anything, to remove all
strain from myself too ;-) (Maybe it'll help to kill the first signs of a
flu I noticed as well.) Sitting around with friends is a good first step,
refraining from hacking on MirLibtool might increase my health again, and
if I'll ever get lunch I'm happy for today.
Later that day I managed
to sit in the Netzladen again – my friends have dragged me. Lacking
anything useful to do I've brought out and debugged (keymap on GNU/Linux)
joe-3.1jupp3.
12.02.2005 by tg@
I took my time, after finishing the work on Jupp
(and building mac and win binary packages), to update PHP 5 to the latest
version and apply the Hardened-PHP patchkit. It's still considered
experimental, however it works on odem. It was a nice evening at Yofuh's,
with us five "core BOSng members"
chilling with Club-Mate (as usual), and we finally got around to some web
site hacking, too. Benny is at another IRC comrade's place, eating sushi,
the first time for him, and seems to have fun playing Counter-Strike (the
Windows® fanboy he is ;-)
13.02.2005 by tg@
After finishing php 5 (except for naming issues)
I took care of Samba 3 and started the conversion to PostgreSQL 8. That's
my first time meeting @sample and luckily benz helped me with it
(thanks). He did a huge docs update (after telling him how worn out I am,
he pointed that out – but who did MirLibtool some days ago
I ask here? ;-)
Now there's left: clean up after Samba investigating
the ProPolice problem as well as the unability to change passwords. Here,
gmane jumps in again – it's funny that most mailing lists are gated
there, most even bi-directionally. I sent out a posting and now idle here
waiting for the answer. Good night!
14.02.2005 by tg@
After work I went into the Netzladen, giving the
visitors CAcert points and finishing the PostgreSQL 8 port. It's a weirdo
bringing its own timezone data, but it looks like it's wanted that way. I
just hope their ohlson-libtz variant correctly handles our native time_t,
varying between 32 and 64 bits. They use an internal pg_time_t of 64 bit,
though. Time to sync the INDEX.
I've made a #machackers IRC
channel for them @ irc6.oftc.net:6667 so they can't throw me out
of the Netzladen tonight ;) Greets to Anglo, Scotty™, ane and johl;
be sure to visit the channel.
20.02.2005 by tg@
I've about recovered from a flu. I've got to get
working tomorrow again, even if I wish I could take the day off, but it's
too hard to get a testate. I've got to take care of a few things the next
two weeks, so don't expect much commits in that time.
The new ports.
Troublesome (to speak with Shikamaru). Samba doesn't work as expected and
I've got no way to test PostgreSQL; pine and mplayer could need an update
(though they can wait until the next version is released).
20.02.2005 by sap@
More news than usual. Got a part-time job to
complement my full time one, and looks like I might have some more work
to fill my weekends in addition to my days and nights. It sounds awful
but I couldn't be happier, it's interesting work and the extra money
doesn't hurt, and I get to arrange my hours around my family life. I
guess that wasn't really too much news, but still more that usual. I
just realized that this is the first wlog entry I've written in two
months. Time flies when you're having fun I guess, or when you don't
have much to say.
20.02.2005 by bsiegert@
The GNOME update is progressing nicely. The only problem is that it is an
awful lot of packages to port. I will probably do a meta port like FreeBSD
has: this port does not contain any code but depends on all packages in the
right order. Such meta ports could also be done for other stuff, e.g. a
complete mail server setup. In any case, I have removed the additional runtime
dependencies for the GNOME packages that should assure that everything is
installed—these never worked right. I also discovered a bug in the
handling of -pthread in MirLibtool that is fixed now (thanks to tg@).
Last weekend, I visited ssimon from our IRC channel—yes, IRL. We had
some fun eating sushi (mmm) and playing games like CounterStrike under Linux
in cedega. I had winter holidays this week, but alas, they are over now. As
usual, I am writing this entry in the train.
We also came up with the idea to port MirPorts to Mac OS X. Work
has not yet started, but it seems like quite a few people are interested in
it. Ironically, the Objective-C extensions to the pkgtools are going to be a
problem because the GNU folks deliberately changed the spelling of function
names in their libobjc. The OS X version however is a "real" NeXTstep
implementation with proper spelling. Sigh.
I also bought a CF-to-IDE adapter which
works fine. The goal is to install MirOS-stable to the CF card (256 MiB) and
use it instead of a hard disk in frankie. I imported flashdist (which is for
OpenBSD) to contrib/code/flashdist in cvs and began playing with it.
Let's see how that works out.
23.02.2005 by tg@
I've been having a birthday party yesterday with
the coder of the first Mac OSX worm ever (he's exactly one year younger).
The party happened in the Netzladen (what did you guess), but ended badly
for me when my guests started smoking after his guests did so, while I've
still not fully recovered from the flu I had over the weekend (tobacco is
a REAL pita to my lungs).
Today I've made, out of being bored, first
steps ([link] MirMake, MirPorts' own
infrastructure and OSdep parts) towards running it on Darwin. I hope that
there's at least a smallish userbase which is willing to help, and shared
libraries work with a bit of sed magic in PFRAG.shared, to happen in make
(and not in the pkg tools, unless bsiegert@ has really good arguments). I
myself won't do this though, since I despise the Macintosh, Motorola (and
PowerPC) and its followings.
24.02.2005 by tg@
I feel left alone at some places, the Netzladen,
among others. I'll continue to fight on all fronts, though.
I played
with MirPorts today, modernising the MODULES= stuff and preparing for the
long-planned Makefile-for-subdir revamp I already did for e.g. the former
gfdl-books port.
I've sort of shut down my brain regarding planning,
preparing and announcing the trip to FOSDEM. Sensory overload, or so. The
BSDs won't have a booth, and I admit I didn't ask them day for day for it
(in the hope they'd give in). I barely will make it for the annual "early
bird gathering" and have to get off on Sunday already. I seem to have got
a place to sleep, too. I've got to work tomorrow (btw, does any of
the English speakers who read this ever use the term "vocational school"?
I don't think I've heard it before), and I've got thousands more things I
need to take care of, such as securing a basic income. My mood varies not
only a bit now, I often switch between happy, energy-laden and depressed,
several times a day, and rarely get the creative mood of yesterday again.
I can only interpret this as "I need a vacation". Other project leaders I
won't name have just had one...
27.02.2005 by bsiegert@
It is that time of the week again, and I am sitting again in the train but
this time from Bruxelles to Strasbourg.
FOSDEM was an adventure, but getting
there and getting back even more so. On Friday, I accidentally took the wrong
train at Bruxelles-Nord, today I almost missed my train because the bus was
much too late and did not stop where I wanted to leave it. I found myself
somewhere in the centre of Bruxelles, but luckily, I discovered that it was
only like 500 meters to the Bruxelles-Luxembourg station, so I catched the
train there.
On Friday, we were at the early-bird gathering at Le Roy d'Espagne,
a bar right at the Grand Place, where tg later joined me. We had some beers
and some talk with the others—KDE, Debian, and Fresco developers. We
took the last train to Leuven and basically went to sleep.
Saturday was the first regular day of FOSDEM. Upon arrival, tg and me just sat
down at the (then empty) Mozilla booth. When the Mozilla folks arrived, they
sent us away, but the organization team officially gave us an empty table with
no chairs as a "booth" after all. With no merchandise, no flyers, and no sign,
it was not a very productive kind of booth, but at least we could sit down and
hack a bit. Later that day, tyler arrived from USA. I went to two GNUstep
talks: the first one was a good general overview, the second one (about
Etoilé) was given by someone who had clearly never done a presentation
before. What's more, he spoke French and had somebody translate it for him to
English. In the evening, we (yofuh, Kabelaffe, cnuke, tyler, tg, and me) went
to La petite planète to eat gyros. After buying lots of liquor
in a supermarket, we returned to Leuven. For tg and me (who took the train),
this took quote long because our train was cancelled and we had to take a
local train instead. tg, cnuke and Kabelaffe went to sleep quite early (after
only one bottle of meade and some beers), while yofuh and me were awake,
drinking red wine and more meade. When we went to sleep, tyler changed the
room to keep on hacking. I think he did not even sleep that night.
On Sunday morning, we went to Bruxelles quite late, and took place in the
"Hacker's room". I went to the Nautilus talk, but left it after two minutes,
after having discovered that the dillo talk was at the same time. The plan9
people were there (including uriel and one Englishman) and kept on trolling
and asking questions like "Why do you use select? Plan9 does not
have it." After the talk, I spoke to Jorge and Sebastian, the main developers.
Jorge was very happy to be offered funding for dillo (maybe). We talked about
a thread-safe resolving library (which BSD does not have out of the box) and
more. Afterwards, I met Wim, who told me to visit the OpenBSD room, Theo de
Raadt would be there. Theo gave a talk about future plans for OpenBSD (for him,
future is after the upcoming 3.7 release) and various other political rants.
Later that day, tg talked to Theo himself and gave him a MirOS T-Shirt. I
would have loved to overhear this conversation but alas, I had to leave for my
train.
During the time, I mostly hacked on the Objective-C stuff in the pkgtools,
which is in my private svn repository for the time being. I have written a
new parser that is able to read and write PLISTs. So the foundations are
there. Stay tuned.
28.02.2005 by tg@
Yay, I'm back in good old Germany, and I love it
(and be it only for having beloved
coffee instead of what they gave us in België). I've published a
few writings, only on German though: at the world-quality German slashdot
clone Symlink:
- Day 0
- Day 1
- Day 2
- A nice followup on this – we missed it even with being there: RMS gives to Theo de Raadt this year's FSF award. Heck, I've nominated him for the last three years in sequence, lucky him.
Oh yeah, and I spoke
to Theo and donated him a MirOS T-Shirt #2. The
shirt was unfolded, and he didn't blink for a second and he asked me, why
I'm doing MirOS and stuff. Keeping up with his spoken English (quite free
of accent, in contrast to the American or British visitors, though) was a
bit difficult to me, and he had to work on stuff, so we didn't talk much;
he did like our work on nroff, though. While on nroff, actually we cannot
use the pre-1989 ditroff since we're not US Americans (our countries have
signed the Berne convention long before). So I've got to either hunt down
BWK, or look at what
the Plan 9 guys have to offer with their ditroff.
Having put several
stickers of FAU-IAA and the Chaos
Communication Club on my laptop, someone gave me a flyer regarding the Anargeek manifesto which fits really nice
into what the Netzladen is all about; thus another post-FOSDEM writing is
at the
Netzladen site. Maybe our local francophones will translate it into a
more sane language ;-)
Success on a large front – the
FOSDEM was quite a chaotic event, with us fighting for a booth and stuff,
but I managed to get ports/editors/joe compiled on cnuke@'s book
running Mac OSX. Now I need to make MirOS paxtar portable (I believe some
other BSD has already done that, though) so "mmake package" can
succeed.
Okay, today I'm losing regarding the length of the wlog, to
Benny – but he had not committed his for 10+ days and added a short
FOSDEM summary (while I wrote a follow-up and did the summary as a series
of Symlink articles). tyler@ got annoyed by some strcpy, etc. warnings in
MirPorts, so he just fixed them out (I'm amazed). He's fun to hang around
with, just like he was in IRC, funny and all, just quite tired. The BOSng
fellows who we usually hang around with, both in the Netzladen and at the
exhibitions, thought so too, and I'm now waiting for the next event to go
to. I'd go to Chemnitz if someone would pay for all my expenses, but I'll
attend Karlsruhe for sure, and Wim invited me to the CCC camp.
03.03.2005 by tg@
I'm sick again. That sucks. I didn't go to work,
but I think I'll make it tomorrow. Wish my nose was freer.
We've got
some more
press coverage (sorry, again German only). Yes, it works now. Jupp of
course, and mc with libglib. cnuke@ lent me his book while kissing (for a
good part of the evening). Now I've got to get my ass up for merging ncvs
1, finally. But bsiegert@ still hasn't done a thing (except bitch around)
for it. Oh, and I don't get the hang on OSI.
ZDnet shot photos of us
at FOSDEM, but they aren't available. So, I guess that's it. I wish I had
let someone take a photo of deraadt and me, but it was kinda quick.
04.03.2005 by bsiegert@
Going home by train is a little chaotic today, because the EuroCity was broken
and I had to leave it in Karlsruhe. I had quite some time in the afternoons
this week, and I used it. After tg's latest bsd.port.subdir.mk
changes, I upgraded my local ports tree checkout and discovered lots of
uncommitted changes, so I committed (some of) them. The GNOME 2.8 migration
in MirPorts continued smoothly. I did some stuff that OpenBSD does not have
myself: libgnomecups (CUPS hooks), libxklavier and libxklavier1 (high-level
xkb API), libcroco (for CSS support in librsvg which is required by the SVG
spec) and CUPS support for GNOME-Print. I could not test this yet, but GNOME
apps should show for example all the CUPS printers in the printer list. OTOH,
I actually had to downgrade some of these because the new versions are already
incompatible. They are for GNOME 2.10 which is going to be out soon
*sigh*.
Anyway, with gnome-session, metacity, and nautilus running, you could already
try to use it. However, I still do not consider the upgrade finished. Not
until gnumeric is running (I need this) and the meta port is there.
05.03.2005 by tg@
You don't believe what a good night of sleep and
a breakfast of bread and tea (no milk, so no coffee :( can make of you. I
still am ill, but I hope that will improve. I'm not, in contrast to Benny
updating GNU GNOME, going to take on KDE. I leave that to Marc Espie, for
he's far superiour in that area, and going to concentrate on base now, in
order to be able to throw out new snapshots soon. I'm going to need them,
for spreading around the word and stuff. And tyler@ needs libstdc++, so I
don't have much of a choice. OpenBSD seems bitten by ProPolice bugs; same
as us but their test programs work fine on 3.2 and 3.4 here.
Later
that day. My dear friend and Mach-Acker Angelo, Coder of the Mac OSX worm
#1, says
it all. These OpenBSE people are so poor...
And: who the fuck is
"Harald Freyer"? (Not having a good opinion on Debian derivate from Canonical that cannot be named
(dubbed "Negergnu" by me) is not necessarily racist, nor does it have, or
was intended with, racist undertone. I'm sick, not being able to say what
I think. Who am I, a German born 1910 or 1982?)
Well, all the strain
aside, it was a nice evening. I liked the fresh air, afterwards we (cnuke
with his gf, as well as Angelo, I and a BOSng companion) had a good meal,
after a long time, at our favourite jugoslawic food dealer again. I think
I need not say I like Julishka more than SSH Connection Multiplexing (and
you know how excited I got over it?).
I wish I could be on the event
Linux-Tage in Karl-Marx-Stadt, but I'm getting over it already. Buttt I'm
really looking forward to our joint booth at the LinuxTag Karlsruhe, with
the Holarse gang.
06.03.2005 by tg@
I've started re-merging ncvs I to ncvs ][ today,
but first I had to discover several import flaws, so I hacked up a bit in
the repository. (I know, I had promised to not do this again. Well, looks
like since we use CVS not SVN, you wouldn't have even noticed if I hadn't
written it here, or you read ancient CVS mails. You won't notice it while
using, promised. And it looks fine now.)
I've merged, so far, all of
src except usr.bin, usr.sbin, gnu/usr.bin and gnu/usr.sbin; the generated
files neither of course. I'm awaiting another Lynx development version in
a few days, and let's hope XFree86™ 4.5.0 comes out before I'm _so_
through with src and gcc and making a new snapshot to start working on X.
There are really good chances, plus I've got to make libstdc++ work first
(and maybe Fortran? Who needs it? gfortran (gcc 4) won't be part of base,
for library reasons, anyway, so?) to please tyler@.
There has been a
libtool update (they realised themselfes (well, SuSE did) that they broke
pthreads), and Benny threw in some Darwin stuff.
My health is nearly
back now, but I've got to work next week too.
Tells that the person,
"Harald Freyer", I asked for yesterday, really is hubertf@nbsd.
I skip commenting on that here, though. Anyway, I'm lucky that MirOS _is_
something special and we'll share a booth with Holarse, and not any other
BSD, at LinuxTag.
Any volunteers outside? While skipping through ca.
6 MiB unified diffs today I noticed our man pages aren't all in sync with
reality, and I still want IPV4_MAPPED addresses
(in IPv6) to work.
We freed the mirbsd.org domain, with the
nice BOSng lad who went eating with us the
day before, re-registering it and pointing it to the same server on which
mirbsd.de runs on, cooperating with Waldi. I've even set up mail
for the two domains now!
Update: look at this Mar 6 18:38 Directory [4]4.4.99.903 Mar 6 20:16 bzip2 [2]XFree86-4.4.99.903.tar.bz2 40507Kb Do my dreams come true or what? I'm eagerly awaiting this, and be it only to be able to /kick from IRC these who mention "the other X11 implementation". ;-)
09.03.2005 by tg@
It's become quite silent recently. I'm relaxing,
listening to some music,
classical one, and shut down my system with +90 instead of now. I think I
deserve a break, especially considering my health, and cnuke@ has taken a
break from all computing related activity as well (now I wish I could hit
the road to the swimming bath, walk the forest with cnuke, do some fires,
and related stuff). At work, my apprenticeship's final project is keeping
me busy, as does the flu (I had fever today at work, the first time since
all this illness fuss started!). mirbsd.org should work as eMail
domain (an alias to mirbsd.org) soon enough. We're continuing until first
lawyer letters regarding patents come in, I think. The OSI is a bunch of,
politic even, wusses who now have made a public mailing list an official!
committee of theirs. (I wonder if I am, now, too?) I started flaming some
on their lists, just for the fun of it (gimme my fun, will you, since Wim
doesn't send me Kriek he owes me for their last pun).
13.03.2005 by tg@
I've finished merging src (except gcc) over from
ncvs1 to ncvs2, except for lynx, where I'm waiting for the new version at
upstream which is due RSN. I haven't looked at XFree86™ yet. Now I
will try to throw in some goodies from OpenBSD-HEAD, then build what I've
got right now. After that, I'll peek at gcc, build what we have and after
that look at libstdc++ (hi tyler) and lynx. If a new XFree86™ comes
out in the meantime, I think I'll have a look at it with higher priority,
since I want to feed back our local diffs, and don't want to run -current
as MirOS default X11 environment.
14.03.2005 by tg@
Today I've built the first kernel from a working
copy of ncvs ][ sources. I've ensured that GENERIC builds as well as i386
machdep (mbrldr, mbrmgr, ffspbr, installboot, boot, pxeboot) stuff. If it
boots I'll see later (won't check mbr/pbr/boot ATM though). After some of
the OpenBSD-current fixes go in, I'll repeat the test, in order to know I
haven't broken anything during a large upcoming merger (mostly userspace,
but one can never be sure, and NTFS, msdosfs and UFS are touched too). It
is a small step for me, already on experimental sources, but a large step
for the whole of MirOS, as gcc3.4 specific constructs, both MI and MirOS'
extras (#pragma), timet64-specifics (working with timet32 @sparc tho) and
other dependencies on MirOS particulars are deeply ingrained now. The new
kernel is also a test if the merger went well, and for the security fixes
and behavioural changes of the last few days. In addition, I have quite a
few files regenerated, possibly breaking preexisting behaviour (damn, and
I always wondered why rsync would not work without its own, bundled libz,
there I go).
Tomorrow I'll have the "KMK exam" – the
teachers over at my vocational college (workschool) explained it to be an
internationally accepted extra qualification (certificate), documenting a
certain level of English knowledge exists. I'm of course going for the #1
level since I've read enough man pages to know the better part of English
I'll need. They'll do lengthy oral exams too, though.
I now have, on
my website, these weird Google ads,
as a kind of beta test for the MirOS pages. On the latter, though, you'll
be able to switch them off by means of a cookie, using MirStyleSelect (as
it provides a preexisting infrastructure quite well). So don't wonder. If
you ask where the money is going, be sure it'll be used for MirBSD®'s
further development and deployment.
15.03.2005 by tg@
W00t, we've finally got a logo for MirOS BSD. On
the other hand, most of you will already know it. Stay tuned, I'll update
the framework as soon as feasible.
I've built a new base running the
kernel I built yesterday from fresh sources. So far, no breakage. Also, I
have added splitb(1) as split(1) improvement (MirOS specific), cleaned up
lots of Makefiles and fixed remaining breakage. The hardest part of that,
but entertaining, is the changes to the -mdoc package for nroff. Finally,
we won't have problems with \&! or ?\& any more
and I've added a workaround for .Xr st 4. Also, .Xr now
temporarily switches to non-hyphenation mode (and back to hyphenated once
it's run, even if it was non-hyphenating before).
make b-r,
except kernels, succeeds now as does make distclean in /usr/src.
I've also updated the packaging lists, so I could quickly make a snapshot
(though incomplete). Much redundant code was removed, and the MirOS brand
applied to many files.
I've also done a licencing cleanup in the www
repository and added a few clauses needed for a new contributors' logo or
mascot component.
Benny has implemented @lib for ports that
has proven superiour for Mac OSX compatibility, so I've deprecated all of
the old DYNLIBDIR() stuff, and Benny removed them. Minor breakage planned
as usual.
16.03.2005 by tg@
Even with nice weather outside nothing restrains
me from freezing ocvs, and so I did. There's however a new seed file that
shows the final version of ocvs to be archived. (This also means: EOL for
MirOS #7-stable.)
I've translated the anargeek manifesto about which
I already spoke into German and designed a logo for it; bsiegert@ has yet
to check it against the french original text.
20.03.2005 by tg@
Okay, even the nice weather hasn't limited me to
update XFree86™ to their latest released version. I was outside for
a while though, and that helped me inspire. If I would now just get these
damn scripts done I need for $ork tomorrow...
21.03.2005 by tg@
A happy start into the spring... XFree86™,
as announced, works. I've imported gcc 3.4.4 (prerelease), but gotta work
today, and every day of this week 'til thundersday. Promising activity is
ongoing by the ppl from OpenBSD again, making everyone realise how unfree
"FreeBSD" really is. I'm quite often astonished when I read the
news, but amazed how high-quality they want to keep their code. I can say
I'd like to follow that, and despite being an amateur project we will try
hard. Everybody can help, and if it's just a donation, a small docs patch
or you buying a T-Shirt. Testing snapshots and reporting broken behaviour
is of course better; helping to fix that best.
We still don't have a
"public relationships manager" and I'm waiting for the new logo
but we'll appear with Holarse and BOSng at LinuxTag, Karlsruhe ;-)
22.03.2005 by tg@
Wow, I did a fire for spring equinox, and now in
my village-like thing they're going to do one for Ostara, although with a
real roman-catholic background ;-)
In the meanwhile, we've received,
edited and added the logo I've promised, and gcc got extended by SSP. I'm
still working over the day so it's going at a slow pace, but is going on;
joe 3.2 is out and I've yet got to have a look at it.
24.03.2005 by bsiegert@
Lots of things happened since my last wlog entry. I finally removed all the
traces of the old DYNLIBDIR code in MirPorts, replacing it with
@lib using a script. MirPorts has made the transition to ncvs2.
During this process, tg renamed lots of ports, and I had to touch a lot of
ports to cure the "missing dependency" errors. I still have not done a
new index. Only after this can the transition be considered finished. I also
fixed a Darwin compatibility bug in the libc hash code, but there is no new
MirMake release that incorporates it yet.
I also made a program for my polymers course and put it under
contrib/code/polymers. However, the second version has a strange bug
that causes the program to go into an endless loop every 2000-9000
invocations. I will have to investigate this further.
26.03.2005 by tg@
People say I've been quite stressy lately. Well,
that may be true, I'm working on my apprenticeship's finishing stuff, too
(and Winword plays its games with me). And I've still got the feeling I'm
talking to a wall, for example when asking for the events of the last 3-4
months.
We (HolaBSDng, that is) have got a booth for the LinuxTag in
Karlsruhe, join us.
gcc 3.4.4 (20050326; prerelease; propolice) eats
up my nerves, but builds, systrace'd and all, I've just left out the java
stuff for now. I'm looking at exception handling through a shared libgcc,
if anyone has got test cases I'd be glad to hear from it.
darkveggy,
the #1 anargeek, has answered, and I've pointed him to what I've drafted,
this page and image
(SVG) I drafted. Let's look what he's got to add.
I've installed
greylisting on the MXes (thor and loki) and miss my spam now ;-)
27.03.2005 by tg@
If I could kill... well, if I cannot get C++ for
us, I will at least try to get Pascal working. But without docs, I am not
going to rewrite 80% of the Makefile to prevent it from placing generated
files into the source directory. Besides, the gpc docs are online I think
(and I could always add them later, differently). Lynx is back and Master
Dickey is helping me to get a current version. I will try if updating the
binutils stuff can help C++ (and we didn't start with gdb anyway).
I
have missed DST getting in effect today again – can't just everyone
use UTC? It's hard for me too to go to bed at 20:00 and wake up at 03:50,
but if I can do it...
28.03.2005 by tg@
I'm running the GNU Pascal testsuite and updated
binutils to a newer CVS snapshot in the hope of fixing C++. Also I try to
work for release engineering...
31.03.2005 by bsiegert@
The ports transition to ncvs still causes some dependency breakage, but I hope
that most of those cases are really fixed now. I am porting
bsd.lib.mk from MirMake to work with the Darwin shared library model.
My testbed is graphics/jpeg which builds libjpeg itself and assorted tools.
The problem is that on Darwin, you don't have s shared library cache that is
updated with ldconfig. When you build a shared library, you have to pass the
install name along with the path, like this:
cc -dynamiclib -install_name /usr/mpkg/lib/libjpeg.62.dylib
-compatibility_version 62.0 -current_version 62.0 -o libjpeg.62.0.dylib
[object file names ...]
Programs that link against this library take the install name and save it in
their executable. But when you use MirMake, you don't know the final install
location (the PREFIX) while building. Thus, we have to relink the
library during installation to incorporate the correct -install_name
parameter. But even then, the dependent programs that have been built at the
same time (in my example, the jpeg tools like rdjpgcom) also need to be
relinked, or they won't find their shared library on runtime and give an error
such as "dyld: rdjpgcom can't open library: libjpeg.62.0.dylib (No such file
or directory, errno = 2)". tg@ has offered to rewrite the Makefiles to use
libtool instead, but I don't think I want that. Another portability problem is
the _gplwarn.c file from the infrastructure which is ELF-specific
and gives syntax error with the Darwin linker *sigh*.
02.04.2005 by tg@
Just in case anyone wonders where I have been at
all the time, yesterday, today. I was outside and enjoyed sun and nature.
Recreation is a powerful measure to refill /dev/random – I
can do it myself and need not ask others like our colleagues of a certain
other BSD. Also, some -funroll-loops freaks have stolen my idea to port a
reasonable Unixish operating system to Interix, but they couldn't finish.
Gets a smile from me.
11.04.2005 by tg@
I've been outside a lot again, since finishing a
documentation (process-oriented project report) about my apprenticeship's
final work got me really under a strain I never experienced before, while
pushing the 2005-03-30 snapshot out of the doors did not help much either
with it.
Today and yesterday, I've started picking up the make diffs
bsiegert@ has left me with for a week and brought them into shape, adding
support for PE targets (MS Interix and MirInterix) and considering other,
non-Darwin, Mach-O targets. I also cleaned up MirPorts appropriately, for
we already had two platforms and two OS kernels in one platform and could
not say a port is only/not for some platform until now; today I've added,
as supplement to OS:VERSION:ARCH, my beloved ksh's wildcards.
All of
today's commits were prepared under Windows® 2000 – in parallel
I almost broke my neck getting the CVS client up and running under almost
pristine MS Interix (for MirPorts, I'll take a peek at pkgsrc but for now
I'm doing this all on my own), in order to compile my fist MirPackages in
Windows®. This port is easier than and inspired by the Darwin port of
MirPorts (well you could even say the MirInterix support for MirPorts was
first, in my head at least, and the Darwin port inspired by that), and no
NetBSD® pkgsrc's developers' pride was hurt in doing so (this is just
to prevent anoter entry in RIPOFF). Eventually, we'll be able to support:
MirOS BSD, MirOS Interix, (MirOS Linux); OpenBSD; vanilla MS Interix/SFU;
Mac OSX, Darwin, OpenDarwin. For old OpenBSD versions and Interix, I must
go the Portage way and throw most of (MirOS) base into ports (but without
removing them from base), because Interix doesn't even have CVS, SSL, SSH
and "tar z", and old OpenBSD lacks XFree86™ 4.5 as well as a modern
GNU CVS 1.12 (well, new OpenBSD lacks them too).
I've been toying at
the GCC testsuite, but I lack experience to parse the non-obvious results
and have no idea what I've been doing. But ports look bright, even though
bsiegert@ is off for a week.
12.04.2005 by tg@
Today, I decided that investing some time in the
MirPorts/Darwin port will be more effective towards my MirPorts/MSInterix
port than directly working on it. Ta-daa! We now have ${.SYSMK} and cksum
for Darwin (as working port!), a new MirMake version, a MirCksum distfile
and a bunch of other goodies.
I decided to add support for sedecimal
number parsing to strtonum(3), much to delight of tyler@.
13.04.2005 by tg@
Of course, my good purpose, to take a shower and
go to sleep early today, didn't become reality – but I am almost at
the break-through point for MirPorts/Interix. Now we only lack ftp(1), or
an equivalent downloader. Thus, MirPorts/Darwin got updated make and cpio
as well (not necessary to update, though).
Oh, nearly forgot: I hope
you don't mind the ads and associated CSS cleanup. I've found they can be
nicely integrated.
14.04.2005 by tg@
I've liberated our version of XFree86® (with
mailing the Xlib GPL compatibility diff upstream, of course), and applied
a bunch of patches from others. I'll rebuild my own system RSN to test it
(wanna bet I didn't break anything?). OpenSSH got updated too, a few more
experimental diffs entered the kernel, and netinet6 was updated in total.
ports/sysutils/pxegrub builds again.
15.04.2005 by tg@
Dang! Instead of going to bed, I seemingly chose
to go into the Netzladen and help Kabelaffe to get MirOS on "xTerm", some
really weird touchscreen machinery of his. I cannot fix the touchscreen's
driver though, it doesn't speak serial so it's a kernel leven thing. Also
we've got a Google search now (please click the ads *hint* *hint* ;-) and
I've rebuilt my entire system (and synched the lists). terminfo(5) is now
correctly installed (it was pregenerated on some OpenBSD system, I should
probably re-do it with a natively built GNU groff, if I had one), but the
htman sets still don't build (so no dist set sync) due to vgrind being in
csh. I've also fixed the N'34' error of XFree86®'s man pages
(which is actually a GNU groff extension, but can be taken care of by our
nrcon -8 or the new -e option if you don't need an 8-bit pass-through); I
have changed X to build its man pages using nrcon -e now.
19.04.2005 by tg@
Benny is back and I have hacked on a whole bunch
of stuff in the meanwhile. Today, it's the GNU Debugger (it bugs me!), of
course with the "I am shitty and restrict you" documentation license(sic)
and a variable called "main", next to 1000 gcc3.4 errors. Should work for
low values of work though, and libedit got improved also (and nbsdified).
If libedit catches up even more, we can have the TUI. More later.
Of
course, security hints – such as the following IRC paste:
[13:06] <benz> [8]http://www.heise.de/newsticker/meldung/58736 <--- mira, für dich
always reach you when you're going to go to bed. Anyway, it's in now, and even has got a nice new feature – we're slowly going for supporting the long awaited change sets.
20.04.2005 by tg@
Yummy pancakes at grandmother's place let me get
back on track; the new snapshot is out
and Benny is back. We've fixed some stuff in ports, and X11 is improving.
I'll spare to comment on the new OpenBSD song lyrics.
24.04.2005 by bsiegert@
My two weeks of holidays are over now *sigh*, and a week of hard university
work lies before me. But Gran Canaria is very enjoyable at this time of the
year, I can tell you. Regarding MirOS, I have worked mostly on
MirPorts/Darwin, but other MirPorts users should benefit as well from my
updates. Now that the make infrastructure and even _gplwarn.c work
(thanks to tg@ for the latter!), hacks like the one from libpng can be
safely removed. Mac OS X users, please redo "make setup"
for the new MirMake version. I discovered some unavailable distfiles
(xloadimage and libtiff), so I had to update the packages. The new libtiff
build system is actually very nice—I just found two minor bugs, the
patches will be submitted upstream.
25.04.2005 by tg@
Yes, I'm making astringency now with the release
and we'd better fix ports and get documentation soon. I'll prepare a FAQ,
with entries such as "Why does it not find <bsd.port.mk>"
for our beloved users.
I've found the g++ bug, finally. I must admit
it's due to gdb. It lured in CSU – the constructors were never run,
due to a change to the include path. I should probably grep all Makefiles
for use of -I in CPPFLAGS (no CFLAGS left, thankfully) and use -idirafter
in some of these places? This is the second time I get bitten by this, if
I'm not totally confused.
26.04.2005 by tg@
Next to some minor ports updates, I also attempt
to build gcj. It mostly succeeds, except for *sigh* some weird C++ file I
can't fix (obviously). Weirder: it builds if I pre-process it first, then
nuke the line number information. centericq works but doesn't convert the
UTF-8 characters to latin1 even if I tell it so.
28.04.2005 by tg@
It's Netzladen time, and since I have an exam on
Tuesday, I don't need to work tomorrow (it's the final exam), so I can be
here long and drink heavily with wbx@ while fixing some ports, leading to
discovering more flaws in base. I've also made an OpenWRT package for our
beloved ksh ( 1 ) although wbx
is mentioned as maintainer – since I don't use it, I can't maintain
it appropriately. Fun work, though. And it clearly shows me just how much
superior MirPorts really is.
30.04.2005 by tg@
Instead of a wlog entry, today I decided sending
a "status"
mail to our general mailing list. I hope this mail assists more users
into helping us. I'm sitting at a friend right now and may get an old box
(laptop with broken keyboard) as build machine.
01.05.2005 by tg@
41 years of BASIC, Boys Day in Japan, Day of the
Queen in the Netherlands, gcj first working in MirOS. Though weird and of
course, boehm-gc prevents gij from being useful (again). But still the 99
bottles of beer are passed around in various ways.
It's hot outside,
too hot to think, but better get a tan outside than to sweat inside.
03.05.2005 by tg@
We've made it through – the IHK exam _was_
pretty easy. I'm sitting in the Netzladen with a few friends, some had to
do the exam too, some just are having a beer or Club-Mate with us. During
the evening, I discovered httpd's brokenness and fixed it. I should write
a bug report about gij, though. All in all, MirOS #8 _will_ rock, I think
– thanks to the help of my co-developers and the support of friends
and foreigners. I've also updated qemu, should probably take on pine, and
abiword, next.
05.05.2005 by tg@
I hate regressions. Why it didn't attach apm0 (I
had added powernowhack0 at mainbus0, preventing bios0 from attaching). An
error on my side, but why the fuck isn't something like this documented?!
Anyway we're almost there. I've already listed most of the required
advertising clauses; the rest will follow soon. The CVS server now runs a
third release candidate (which I unfortunately cannot upload, due to thor
not being reachable by any means including serial console), and should be
faster because of improvements in the scripts. We've got a preliminary FAQ with lotsa content to follow (I always get the ideas
on the toilet, though, where I don't have a shell). I'm tired, should put
more here, but hey, at least I'm writing again. As usual, read the commit
mails (or ChangeLog files) if you want to know more.
12.05.2005 by bsiegert@
The #8 release is going to be Really Soon Now™. We are somewhere in
the release candidates—RC2 or RC3, I don't know. It meets at least
the goals I had for a release: both gdb and g++ work now, and even Java
seems to be unbroken, both native (gcj) and the Java VM. I have started
writing an installation manual in troff (mdoc) which will be the
install.i386(7) manpage. It contains some theoretical chapters about
partitioning etc. and a walkthrough through the MirOS install. While
writing, I had a qemu session with the installation in parallel. However,
I couldn't figure out how to change CDs while the emulation is running.
Thus, I did not get to installing the sets, having booted from the
Mini-ISO :(.
Some things in MirPorts were fixed as well, e.g. devel/pcre which
actually contained a pre-MirLibtool patch that broke the build. tg@ found
a bug in the pkgtools (introduced by me) that made them generate wrong
library dependencies. @sample now tells you both filenames when
you need to update a configuration file yourself. I started an update of
firefox to 1.0.3 but it is a lot of work.
I have also finally backported the OpenBSD security patches from about a
month ago and applied them to #7-stable, so a new upgrade is going to come
soon. But alas, alas, I left my workstation in Strasbourg switched off
(with both the unfinished installation docs and the uncommitted #7 patches
on the hard disk), so I cannot access the data at home over Pentecost
(sp?). But still, I wish you all ein gesegnetes Pfingstfest.
19.05.2005 by tg@
I'm a bit lost in all the spring mood but I need
a distraction from all the hacking, at the dayjob and in the evening. The
changes I've done are numerous, small and large; I know I'll kick myself,
later, when trying to collect new snapshot data. When trying to upload RC
3 (via rsync) thor went wild, but Strato fixed that some days(!) later; I
haven't uploaded it tho. Today I've vamped mplayer and fixed these gcc or
boehm-gc bugs (some of which were introduced via threading gcc itself and
thus infecting libgcc). Maybe I'll start merging binutils/gdb-current now
since I'm not yet in the mood to drive home (Netzladen time again). We'll
have an IRC channel meeting this weekend, and I'm learning dutch.
23.05.2005 by tg@
I just discovered I had missed to 'cvs add' that
FAQ sample I wrote some time ago... done. While there, I've crafted a new
mirbsdksh, currently in beta testing by myself on MirOS, not amended with
the portable stuff yet, but smaller and more featureful. MirPorts/Interix
have become usable except for the lack of a downloader; with ${FETCH_CMD}
set to runwin32 C:/usr/local/wbin/wget -c in /etc/mk.conf I have
been able to build at least some of the stuff, awaiting further pkgtools,
infrastructure and PLIST changes by bsiegert@.
26.05.2005 by tg@
Somehow I don't get to work on the areas where I
really should work – binutils, gdb, gcc, looking for da advertising
clauses, lynx. But we've got a damn well working newish mirbsdksh, and on
some webpages we have been mentioned... Linux from Scratch for example as
ProPolice is being heavily used by us, making us suitable for hosting the
NetBSD cross build environment.
28.05.2005 by tg@
As you can see on the MirbsdKsh page in the Wiki
I have created a Debian source package, and Han Boetes has made a port to
Crux GNU/Linux (not that I didn't have one either...) of mksh R22c. Maybe
it will be in Debian 3.2 in ten years or so ;) (j/k)
I have upgraded
the in-tree gcc to the first 3.4.5-prerelease (latest 3.4-stable) version
and will continue to work on GNU binutils later in order to make the tree
buildable again. Benny is slightly busy with port updates and creation of
new ones, and we've both dived a bit more into the particulars of Interix
and Darwin.
We may produce a release song ourselves...
07.06.2005 by tg@
Nothing important happening; some binutils merge
actions started, but not finished. Stay sleeping for a week or so.
09.06.2005 by tg@
What a surprise; with new mirbsdksh and mirmake,
stuff works _way_ better on Interix, out of a sudden. Expect some further
improvements in that department soon.
I have my final examination on
June 20th; so far, I've been doing well, scoring 100% in some part of the
written qualification. They seem to have ignored a whole chapter that was
part of the documentation, though. Linuxtag is on June 21st and following
so I'll be there (donations for costs are welcome, as usual).
As for
MirOS #8 – some sudden lack in interest and time tells me that it's
probably not out there in time, but you can buy release candidats, burnt,
just in time, on CD-Rs there, as usual. There are also still some Shirts.
Thanks to the NetBSD® pkgsrc® (yeah, it's a trademark, too,
now) guys, we also have better Interix support in MirLibtool (ports only,
as of the moment – still testing). The OpenBSD guys, and especially
Han Boetes, have provided us with a
bunch of fixes and improvements and ideas for mksh.
22.06.2005 by tg@
After finishing apprenticeship on Monday, Summer
finally started yesterday – in Karlsruhe, for me. I'm here as usual
for the LinuxTag, currently trying hard to get something to put on the CD
and sell. I've already had two people asking for CDs (not much today tho)
and sold an actual MirShirt to a friendly H+BEDV Antivir man. Also, I had
a nice chat with the GeNUA people again which too remembered me from last
year. LinuxTag is the usual fun with drunkards (not the OpenBSD guys, for
me, but rather the Holarse crowd – although Wim did look funny when
I talked a bit to him op nederlands; had to deliver greets from Han).
26.06.2005 by tg@
LinuxTag is over, but the strain isn't. Some new
users, but less sales than ever. And lots of stuff todo. We will probably
appear on TV.
03.07.2005 + 6 hours, by tg@
Next to shifting off MirLibtool diffs
into GNU Libtool, we also have (probably) USB 2 support and a much faster
system (overall and console) tonight. I have merged about everything from
the OpenBSD-current kernel which I found easily doable (just another hack
session without counting the hours, for my amateurish Frankenstein OS) or
even needed. The userland portions will follow later.
06.07.2005 by tg@
Okay, some people say that USB 2 works. Just the
system's much faster and more reliable to me. Some more work will follow,
then the userland. In the meanwhile, I'm playing around with libtool, and
the gcc module. I just can't bring myself to like symlinks in /usr/lib at
all.
Software patents fight is over, we did not win but did not lose
either. Does anyone know if IDEA is still patented in Europe, then? Or do
we have to fight every single illegal patent out of the 30'000 which got,
wrongly, issued then? Anyone knows of a good lawyer who would work for us
without pay?
MirPorts are undergoing some grave changes (a flag day)
so don't update. benz should keep you informed on this.
07.07.2005 by tg@
As a side effect, we also have SATA support (for
those who think they need it, I go with SCSI not cheap crap), and I found
a bug in mksh(1) which was introduced by one of the Debian patches; broke
<bsd.subdir.mk> in an unfixable way.
09.08.2005 by tg@
I've started auto-converting the JuppEditor, and
MirbsdKsh, pages from the ex-MirWiki, because nobody else did.
03.10.2005 by tg@
Why don't people just fucking use our nice ports
framework on OpenBSD, instead take hours to port stuff to the perl-rotten
OpenBSD ports tree (¾ hour alone to find libpng *sigh*)? The folks
I rant about are updating ports "as they see them" anyway, so there's not
a single reason to not use often-broken MirPorts since the Framework ipso
is now in a stable state, and we'll notify prior to further breakage.
Good luck, have fun, take care etc.pp