On Wed, Sep 24, 2008 at 8:45 PM, Dan Kegel dank@kegel.com wrote:
You wrote:
Fixes a compiler bug when using gcc-4.1
Wait- which is it, uninitialized variables (as your subject line said), or a compiler bug (as your message said)?
Didn't mean to send this one yet, was saving it as a draft until investigated further, must've clicked send by mistake. Ignore please.
If it's a compiler bug, which one? Probably http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24639, right?
Looks like it.
Would setting the fields to any random value work as well?
Hmm, possibly. I'll have to investigate more, as I'm not at that machine ATM.
Can we instead tell the compiler to avoid the bug? (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9049 says there's no way yet to turn warnings on and off for individual functions, sadly, but we can do it for individual files.)
Good to know.
A link to the bug report at gcc.gnu.org, and/or a wine wiki page about the general issue, would be helpful.
I'll probably do something along those lines soon. I've been looking into various GCC CFLAGS/optimizations and have been testing the source code in various configurations. I'll try to get a wiki going of compiler bugs/WONTFIX's, etc. But first, I've got to learn more about what CFLAGS/optimizations/etc. conflict with each other.
- Austin