[] Die, Gecko, Die!
As some of you may know, recently we (ChatZilla) are being screwed
by two bugs in Mozilla’s Core:
- Bug 319219 alias Bug 318419 alias Bug 291386: timers cause your commands and messages to get lost, or get you kicked off the server because we try and use ping / who / ison commands multiple times a second. Not like we want to, but since timers in Mozilla are somewhat broken of late…
- No bug filed yet. For some reason, Gecko thinks it’s funny to give some css rules line numbers like 18234154 - from a css file with less than a 1000 lines, that is. Of course, once it starts giving out random line numbers, the rules get applied in the wrong order, and the cz output document looks horrible. Yay. Of course, this screams ‘minimal testcase’ more than anything else, but here’s the catch(es):
- No matter how hard we try, not everyone can reproduce this
- There doesn’t seem to be any reliable pattern in when it screws up. It just does, at times
- Even a ‘minimal’ testcase would still involve any amount of: a chrome xul document, with a <browser> with type=chrome, loading a chrome:// url html document, modifying that using the DOM to include a chrome:// url css file, which @imports another chrome:// url css file. We know for a fact that all these problems didn’t exist until we set the browser to type chrome to cope with other Firefox stuff (damn you, xpcnativewrappers). Why type=”chrome” and the CSS Parser can’t go through one door together, couldn’t tell you. I’d ask bz, but he’s on holiday for now, so we’re just screwed unless and until someone either figures out what causes this, makes a minimal testcase, or, heaven forbid, has steps to reproduce which mean everyone can actually see the bug…
Edit: Now bug 321195
Sigh. So, Gecko, die.
Incidentally, as if this wasn’t enough yet, my exams suck, and I have no time to do work on CZ at the moment. I wonder if the ~8 days before new year will be enough. Time will tell 