Issue1735030
Created on 2007-06-11 14:06 by mbanck, last changed 2007-07-12 08:13 by georg.brandl.
|
msg52751 - (view) |
Author: Michael Banck (mbanck) |
Date: 2007-06-11 14:06 |
|
This patch (by Wouter van Heyst) fixes compilation of python2.5 on the Hurd (and other systems?)
It fixes the following build error:
gcc -fPIC -fno-strict-aliasing -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -I. -I/build/buildd/python2.5-2.5/./Include -I../Include -I. -I/build/buildd/python2.5-2.5/Include -I/build/buildd/python2.5-2.5/build-static -c /build/buildd/python2.5-2.5/Modules/selectmodule.c -o build/temp.gnu-0.3-i686-AT386-2.5/build/buildd/python2.5-2.5/Modules/selectmodule.o
/build/buildd/python2.5-2.5/Modules/selectmodule.c:676: error: 'select_poll' undeclared here (not in a function)
/build/buildd/python2.5-2.5/Modules/selectmodule.c:676: error: 'poll_doc' undeclared here (not in a function)
/build/buildd/python2.5-2.5/Modules/selectmodule.c: In function 'initselect':
/build/buildd/python2.5-2.5/Modules/selectmodule.c:709: error: 'poll_Type' undeclared (first use in this function)
/build/buildd/python2.5-2.5/Modules/selectmodule.c:709: error: (Each undeclared identifier is reported only once
/build/buildd/python2.5-2.5/Modules/selectmodule.c:709: error: for each function it appears in.)
/build/buildd/python2.5-2.5/Modules/selectmodule.c:709: error: request for member 'ob_type' in something not a structure or union
/build/buildd/python2.5-2.5/Modules/selectmodule.c:709: warning: statement with no effect
We hope this does not break other platforms, does this look OK?
|
|
msg52752 - (view) |
Author: Neal Norwitz (nnorwitz) |
Date: 2007-06-12 02:54 |
|
Is this only change required for Hurd? I'm not sure, but HAVE_BROKEN_POLL might have a specific meaning on OS X. So there might be a problem with this patch. I'd need to look at the whole file to verify. I'll try to remember to do that later.
|
|
msg52753 - (view) |
Author: Michael Banck (mbanck) |
Date: 2007-06-12 12:55 |
|
Yes, python2.5 builds fine otherwise (at least the Debian package, I am not sure whether we applied local changes earlier). This HAVE_BROKEN_POLL issue was the only regression from python2.4 as far as the Debian GNU/Hurd port was concerned.
|
|
msg52754 - (view) |
Author: Michael Banck (mbanck) |
Date: 2007-06-28 16:09 |
|
Hrm, upon further analysis, it looks likely that poll() really is not conforming to POSIX on GNU/Hurd and should get fixed, rather than HAVE_BROKEN_POLL code be changed.
I have submitted this upstream to the Hurd developers, the patch should not be applied in the meantime (and might be obsolete).
|
|
msg52755 - (view) |
Author: Georg Brandl (georg.brandl) |
Date: 2007-07-12 08:13 |
|
Closing for now. If it's really Python's fault, you can reopen this.
|
|
| Date |
User |
Action |
Args |
| 2007-06-11 14:06:55 | mbanck | create | |
|