Summary: | xemacs doesn't compile against linux-headers-2.6.10 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Derek Dolney <z23> |
Component: | Current packages | Assignee: | XEmacs team <xemacs> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | charles, plasmaroo |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | All | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Derek Dolney
2005-02-20 08:39:20 UTC
Ok, I'm not going to fix this in the headers as what XEmacs is doing is Wrong <TM>; doing #define register '' is inherently bad as "register" is a C primitive which headers should be able to fully expect as should application code. Some testing and yanking out the #defines in the XEmacs source is the way to fix this, IMO. XEmacs team, if you believe that this isn't something that you're willing to patch locally then feel free to shoot this off upstream. Is there any movement on this? It's still broken, and linux-headers-2.6.11 is the current ~x86 that gets pulled in. IIRC, it didn't break on 2.6.8 headers (whether xemacs is brain-dead or not...) Try adding: #ifdef register #undef register #endif to /usr/include/asm/thread_info.h... Let me know how that goes. Thanks. I'm experiencing the same bug, I'm trying the fix in #3, will report back momentarily. I can confirm, the fix in #3 does solve this bug. What XEmacs is doing upstream is Bad Voodoo (tm), but the workaround solves the problem for the time being. The compiled XEmacs also appears to work, as best I can tell. Headers patched in CVS, not closing bug as this needs sending upstream. I got this from plasmaroo on irc: 14:51 <@plasmaroo> corsair: Look at 82690 for that Xemacs thing, do the same workaround for paca.h 14:51 <@plasmaroo> corsair: Basically, upstream decides to screw all ANSI C standards and do #define register which is bad. 14:53 <@plasmaroo> corsair: So actually try putting the workaround in the bottom of types.h instead, that might be better. I'll try that ASAP. sorry.. wrong bug.. that should have gone into bug #89770 :-/ This bug is actually fixed in 2.6.11 so I'm closing it; but if somebody would be able to inform upstream about this it would be appreciated. Thanks! |