classification
Title: request to reopen Issue837046 - pyport.h redeclares gethostname() if SOLARIS is defined
Type: compile error Stage:
Components: Build Versions: Python 3.3, Python 2.7
process
Status: open Resolution:
Dependencies: Superseder:
Assigned To: Nosy List: jbeck, jcea, python-dev, r.david.murray, risto3, wiz
Priority: normal Keywords: patch

Created on 2013-11-12 14:39 by risto3, last changed 2015-04-09 01:33 by jbeck.

Files
File name Uploaded Description Edit
Python26-10-gethostname.patch risto3, 2013-11-12 14:39
tp.cpp risto3, 2013-11-13 04:36 test program
Messages (6)
msg202690 - (view) Author: Richard PALO (risto3) Date: 2013-11-12 14:39
I'd like to have reopened this previous issue as it is still very much the case.

I believe as well that the common distros (I can easily verify OpenIndiana and OmniOS) patch it out (patch file attached). 
Upstream/oracle/userland-gate seems to as well.

It is time to retire this check, or at least take into consideration the parametrization from unistd.h:

#if defined(_XPG4_2)
extern int gethostname(char *, size_t);
#elif  !defined(__XOPEN_OR_POSIX) || defined(__EXTENSIONS__)
extern int gethostname(char *, int);
#endif
msg202715 - (view) Author: R. David Murray (r.david.murray) * (Python committer) Date: 2013-11-12 19:22
We have an OpenIndian buildbot that compiles from our source, so no distro changes.  Can you sort out why this isn't a problem on the buildbut but is for you?  Your patch files says 2.6, so is it possible it is fixed in 2.7?
msg202718 - (view) Author: Richard PALO (risto3) Date: 2013-11-12 20:42
I don't believe the problem is a question solely of building the python sources, but also certain dependent application sources... 

I know of at least libreoffice building against python and this problem has come up. 

The workaround was to apply the patch indicated (here, on python3.3).
msg202723 - (view) Author: R. David Murray (r.david.murray) * (Python committer) Date: 2013-11-12 21:56
Libreoffice is a big thing to compile as a test...do you know of any smaller packages that exhibit the bug?

I've added jcea to nosy, he set up the OpenIndiana buildbots and may be able to help clarify the issue.  I'm not willing to opine on it since I don't know what effect modifying those directives might have on other platforms.
msg202730 - (view) Author: Richard PALO (risto3) Date: 2013-11-13 04:36
Sure, attached is a simple test found on the internet, compiled with the following reproduces the problem:

richard@devzone:~/src$ /opt/local/gcc48/bin/g++ -o tp tp.cpp  -DSOLARIS -I/opt/local/include/python2.7 -L/opt/local/lib -lpython2.7 
In file included from /opt/local/include/python2.7/Python.h:58:0,
                 from tp.cpp:1:
/opt/local/include/python2.7/pyport.h:645:35: error: declaration of C function 'int gethostname(char*, int)' conflicts with
 extern int gethostname(char *, int);
                                   ^
In file included from /opt/local/include/python2.7/Python.h:44:0,
                 from tp.cpp:1:
/usr/include/unistd.h:351:12: error: previous declaration 'int gethostname(char*, uint_t)' here
 extern int gethostname(char *, size_t);
msg240296 - (view) Author: John Beck (jbeck) Date: 2015-04-09 01:33
We (Solaris engineering) have hit this issue after migrating from 2.6
being our default version of Python to 2.7 being the default.  The
specific component that broke was vim (version 7.4), trying to compile
if_python.c:

"/usr/include/python2.7/pyport.h", line 645: identifier redeclared:
    gethostname                                                                          
        current : function(pointer to char, int) returning int                       
        previous: function(pointer to char, unsigned long) returning int :           
"/usr/include/unistd.h", line 412                                                    

We had this patched out in Python 2.6's Include/pyport.h:

-#ifdef SOLARIS
-/* Unchecked */
-extern int gethostname(char *, int);
-#endif

but for some reason that patch was not propagated to our 2.7 line.
Until today, that is; I will be applying that patch shortly to both
2.7 and 3.4.
History
Date User Action Args
2015-04-09 01:33:09jbecksetnosy: + jbeck
messages: + msg240296
2014-06-11 06:42:45wizsetnosy: + wiz
2014-03-11 18:06:10r.david.murraysetmessages: - msg213157
2014-03-11 18:05:22python-devsetnosy: + python-dev
messages: + msg213157
2013-11-13 04:36:22risto3setfiles: + tp.cpp

messages: + msg202730
2013-11-12 21:56:44r.david.murraysetassignee: jcea ->
2013-11-12 21:56:34r.david.murraysetassignee: jcea

messages: + msg202723
nosy: + jcea
2013-11-12 20:42:13risto3setmessages: + msg202718
2013-11-12 19:22:40r.david.murraysetnosy: + r.david.murray
messages: + msg202715
2013-11-12 14:39:41risto3create