Bug 208707

Summary:	distutils.sysconfig.{get_config_h_filename,parse_config_h} broken
Product:	Red Hat Enterprise Linux 5	Reporter:	Axel Thimm <Axel.Thimm>
Component:	python	Assignee:	Jeremy Katz <katzj>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Brock Organ <borgan>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	5.0	CC:	katzj
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:	5	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2007-03-26 18:38:49 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:	201434
Bug Blocks:

Description Axel Thimm 2006-09-30 15:24:04 UTC

+++ This bug was initially created as a clone of Bug #201434 +++

Description of problem:
/usr/include/python2.4/pyconfig.h is now a wrapper for 32/64 bit sub-includes.
This breaks the usual query mechanism for extracting python build flags.

Version-Release number of selected component (if applicable):
python-2.4.3-14

How reproducible:
always

Steps to Reproduce:
1.config_h = sysconfig.get_config_h_filename()
2.config_h_vars = sysconfig.parse_config_h(open(config_h))
3.
  
Actual results:
{}

Expected results:
{'STRICT_SYSV_CURSES': "/* Don't use ncurses extensions */", 'HAVE_SELECT': 1,
'HAVE_ST_BLOCKS': 1, [...]

Additional info:
o Breaks builds of smart package
o One can argue that the bug is in the implementation of parse_config_h, which
  should be able to descend into further included files. OTOH it would then need
  to understand nested conditionals, too.
o maybe the 32/64 bits switch should be performed already on 
  get_config_h_filename level? E.g. get_config_h_filename should return
  pyconfig-32.h and pyconfig-64.h respectively?

Thanks!

-- Additional comment from Axel.Thimm on 2006-08-05 06:07 EST --
Turns out this is due to the fix for multilib -devel support in bug #139911
comment 2. This fix only considered C/C++ consumers of pyconfig.h. python's
querying mechanism needs to be redirected, too, by patching
get_config_h_filename. A simple untested specfile fix is in bug #139911 comment 4.

-- Additional comment from misa on 2006-08-17 18:18 EST --
Fix (based on your suggestion) included in python-2.4.3-15 - thanks!

Comment 1 Jeremy Katz 2006-10-01 22:36:56 UTC

Did the fix in -15 not work?  Beta1 was frozen a while back and thus doesn't
include a lot of fixes which have occurred in the meantime; beta2 will
reconverge with FC6 and all the fixes in between

Comment 2 Axel Thimm 2006-10-01 23:45:55 UTC

-15 on FC6 was OK, I cloned the report because beta1 carries -14 and some builds
against beta1 started failing.

Comment 3 Mihai Ibanescu 2006-10-02 02:28:56 UTC

The fix was not a beta1 blocker, therefore it didn't make it. Rest assured it
will be there in beta2. I'll leave the bug open so we can make sure that will
happen.

Comment 4 Axel Thimm 2007-02-07 13:52:08 UTC

I verified that this is fixed in beta2, thanks!

But I cannot close this bug, it says:
 "You tried to change the Status field from MODIFIED to CLOSED, but only the owner
  or submitter of the bug, or a sufficiently empowered user, may change that
  field."

Well, I'm the submitter, but anyway, please close this bug whoever is
"sufficiently empowered". :)

Comment 5 David Lawrence 2007-03-26 18:38:49 UTC

Closing per submitter's request.