Today, I was sort of bored, so I enhanced something other than our libc, namely pcc, with support for Unicode/UTF and wide character literals and strings. Kind of fun it was.
Still missing is support for \uXXXX literals in multibyte strings (need ucs2utf function for that, which can be taken from our libc again), and, maybe, automatic UTF-16/CESU-8 conversions, if the target has UCS-2 (MirBSD, Windows) and the host doesn’t, or so.
Every single time Benny works on ports/www/firesomething, it sinks in my esteem. Generating crypto keys during fake install is not how things should be done (use a post-pkg_add script!). And I still think it’s only rumours that it had been rewritten after the Netscape® code was opened – it would be better than before then, not worse. I’d rather shoot myself than publish such filth, if, like Benny says, the code had indeed been rewritten. Well, at least it is aware how insecure it is and kills itself:
14:46⎜«benz» Nov 23 16:40:29 mirbulk firesomething-bin: stack overflow in function virtual nsresult nsCryptoHash::Finish(PRBool, nsACString_internal&)
And if he wouldn’t document this in a wlog entry, I do ☻☺
But then, I don’t believe him totally unaware:
14:09⎜«benz» The name "Minefield" is strangely appropriate