pyport.h from python2.7 sometimes clashes with c++ functions on OS X

[kiwifb]

The recent upgrade to pynac-0.7.3 brought a flurry of compiling problem on OS X with the following kind of message

  CXX      libpynac_la-sum.lo
In file included from /Users/palmieri/Desktop/Sage_stuff/sage_builds/TESTING/sage-7.5.rc2/local/include/c++/5.4.0/bits/basic_ios.h:37:0,
                 from /Users/palmieri/Desktop/Sage_stuff/sage_builds/TESTING/sage-7.5.rc2/local/include/c++/5.4.0/ios:44,
                 from /Users/palmieri/Desktop/Sage_stuff/sage_builds/TESTING/sage-7.5.rc2/local/include/c++/5.4.0/ostream:38,
                 from /Users/palmieri/Desktop/Sage_stuff/sage_builds/TESTING/sage-7.5.rc2/local/include/c++/5.4.0/iostream:39,
                 from class_info.h:31,
                 from registrar.h:30,
                 from basic.h:40,
                 from numeric.h:49,
                 from inifcns.h:26,
                 from sum.cpp:13:
/Users/palmieri/Desktop/Sage_stuff/sage_builds/TESTING/sage-7.5.rc2/local/include/c++/5.4.0/bits/locale_facets.h:247:53: error: macro "toupper" passed 2 arguments, but takes just 1
       toupper(char_type *__lo, const char_type* __hi) const

This is caused by definition of macros for toupper and other functions in pyport.h specifically on OS X. This is unsafe with C++ and the issue has been known in MacPort for years: https://trac.macports.org/ticket/44288.

We should adopt their patch for python.

Component: packages: standard

Branch/Commit: u/fbissey/pyport @ 502aff0

Reviewer: François Bissey

Issue created by migration from https://trac.sagemath.org/ticket/22147

[kiwifb]

Author: François Bissey

[kiwifb]

Branch: u/fbissey/pyport

[kiwifb]

comment:1

Python3 should be checked over as well.

---

New commits:

502aff0

Adopt macport patch for pyport interference with C++ on OS X trac #22147

[kiwifb]

Commit: 502aff0

[kiwifb]

comment:2

I note that the upstream python issue http://bugs.python.org/issue10910 is marked closed while there was never really a good solution in my opinion.
Anyone can do something about this issue? Can it be re-opened, or do we need a new one.

[jdemeyer]

comment:3

The Python issue was fixed in Python 2.7.13, so could it be that #22037 also fixes the problem?

[kiwifb]

comment:4

Replying to @jdemeyer:

The Python issue was fixed in Python 2.7.13, so could it be that #22037 also fixes the problem?

Depends what's in pyport.h, I cannot see any commit being linked to issue 10910. I can have a look at #22037 which would then be a preferred option to this ticket, but I am running out of time.

[kiwifb]

comment:5

I just checked the python 2.7.13 tarball and the fix is in. So yes #22037 will fix the issue.

[kiwifb]

comment:6

I won't be able to deal with this since I'll be off for a week. I am ok for this ticket to be marked duplicate and #22037 to be marked blocker instead. I see it is needing review so I am guessing it is good to go.

[jdemeyer]

Reviewer: François Bissey

[jdemeyer]

Changed author from François Bissey to none

[vbraun]

comment:8

Still don't understand why it works on the OSX buildbot, oh well...