Needed for bug 232182
Are there any cases where we wouldn't want this defined?
Who "we"? Makefile.ref shouldn't, but README.html should mention the new compile-time option. /be
Created attachment 202343 [details] [diff] [review] v1.0 I would assume "we" means the gecko-based products since this is filed against Core:Build Config. It seems like you're asking for a simple -DJS_STRINGS_ARE_UTF8 to be added unconditionally to the gecko build like we have for JS_THREADSAFE.
Comment on attachment 202343 [details] [diff] [review] v1.0 Presuming from the commentary above that this has moa=brendan, sure.
The patch has been checked in.
I backed out rev 1.1547 of configure.in till we have our heads screwed back on straight, and have adjusted JS API clients that need deflation to ISO-Latin-1 to avoid getting UTF-8. My fault for not coordinating harder. /be
Brendan: Please tell me, what problem do we have? The patch has checked-in to Trunk. But you was backing-out it. Does SeaMonkey or Firefox or Thunderbird have a problem? Or third party's product? If it's latter, I think that we cannot understand the reason of backing-out. Otherwise, our product has some problem, should we use the patch for we fix all regressions?
Please see bug 316178 comment 1. This has caused a very fundamental problem with XPConnect's conversion code.
thanks for the pointer.
I don't see a way we can fix this problem, really -- we have tons of interfaces that abuse |string| and expect to somehow "work" for random data. :(
xpconnect could convert using some other method that keeps the current behaviour.
This should get attention during 1.9. /be
If this is too much for 1.9, it's definitely a Mozilla 2 to-do item, because we can and will break API compat. /be
So much changed around encoding of js strings, the define doesn't exist anymore, I'll just resolve this WORKSFORME. Haven't heard of any of the issues mentioned here in the past few years, too. Feel free to reopen if there's still a problem, though at this point, a new bug is probably better.