Bug 404505 – Python 2.7.2 uses C99 'round' symbol

• Status: RESOLVED UPSTREAM
• Product: Gentoo/Alt

[Description Richard Yao (RETIRED) 2012-02-19 11:25:35 UTC]

Python 2.7.2 uses the C99 'round' symbol in its math module. Portage relies on the math module, which makes it impossible to hack the bootstrap-prefix.sh script in a way that enables the bootstrap process to succeed on Solaris 9 and earlier.

There is an explanation of this at the following page:

https://forums.oracle.com/forums/thread.jspa?threadID=1920897

The exact failure message from the bootstrap process is:

*** WARNING: renaming "math" since importing it failed: ld.so.1: python: fatal: relocation error: file build/lib.solaris-2.9-sun4u-2.7/math.so: symbol round: referenced symbol not found

There is a patch available for this issue at the upstream bug tracker:

http://bugs.python.org/issue9742

[Comment 1 Richard Yao (RETIRED) 2012-02-19 11:34:29 UTC]

Actually, I was wrong about not being able to hack around it. Adding the following line before the configure script is executed will fix the math module failure:

wget -O - "http://bugs.python.org/file20339/Python-2.7.patch" | patch -p1 -d ${S} || exit 1

With that said, there are also failures in _multiprocessing binascii zlib, but I appended them to PYTHON_DISABLE_MODULES to try to get the initial phase of the bootstrap process working.

[Comment 2 Richard Yao (RETIRED) 2012-02-19 11:54:57 UTC]

Actually, ignore what I said. That patch depends on GNU ld, which does not fix this issue:

/usr/local/bin/gcc   -o python \
                Modules/python.o \
                -Wl,--whole-archive libpython2.7.a -Wl,--no-whole-archive \
                        -lresolv -lsocket -lnsl -lrt -ldl  -lpthread   -lm  
/usr/ccs/bin/ld: illegal option -- -
/usr/ccs/bin/ld: illegal option -- w
/usr/ccs/bin/ld: illegal option -- -
/usr/ccs/bin/ld: illegal option -- n
ld: warning: option -o appears more than once, first setting taken
usage: ld

[Comment 3 Fabian Groffen 2012-02-19 12:04:08 UTC]

yeah, so http://bugs.python.org/issue9742#msg134222 seems a bit more viable approach

[Comment 4 Richard Yao (RETIRED) 2012-02-20 22:51:39 UTC]

(In reply to comment #3)
> yeah, so http://bugs.python.org/issue9742#msg134222 seems a bit more viable
> approach

I tried that approach and it did not appear to work. :/

[Comment 5 Fabian Groffen 2012-02-21 07:41:46 UTC]

hmmm, ok, so we need to figure out why the math symbols aren't referenced

you mention you disabled zlib as well, that's going to cause trouble afterwards IIRC

[Comment 6 Fabian Groffen 2014-01-13 21:12:22 UTC]

this should be fixed in latest pythons now