Back-patch support for python 2.5?

I noticed that two of the buildfarm members are failing the 8.1 branch
because they're trying to build plpython with python 2.5. To get that
to work, I think we'd need to back-patch these two fixes:
http://archives.postgresql.org/pgsql-committers/2006-11/msg00165.php
http://archives.postgresql.org/pgsql-committers/2007-04/msg00043.php
Is that worth doing? Should we go further back than 8.1?

regards, tom lane

Tom Lane wrote:
> I noticed that two of the buildfarm members are failing the 8.1
> branch because they're trying to build plpython with python 2.5. To
> get that to work, I think we'd need to back-patch these two fixes:
> http://archives.postgresql.org/pgsql-committers/2006-11/msg00165.php
> http://archives.postgresql.org/pgsql-committers/2007-04/msg00043.php
> Is that worth doing? Should we go further back than 8.1?

I find these patches a bit excessive for backporting, especially 10
minor releases into the branch.

Those who need Python 2.5 can use PostgreSQL 8.2. I consider that to be
a new feature. The fact that the buildfarm exercises other
combinations isn't evidence of a practical need.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> Tom Lane wrote:
>> I noticed that two of the buildfarm members are failing the 8.1
>> branch because they're trying to build plpython with python 2.5. To
>> get that to work, I think we'd need to back-patch these two fixes:
>> http://archives.postgresql.org/pgsql-committers/2006-11/msg00165.php
>> http://archives.postgresql.org/pgsql-committers/2007-04/msg00043.php
>> Is that worth doing? Should we go further back than 8.1?

> Those who need Python 2.5 can use PostgreSQL 8.2.

Not being a Python user, I don't really care. I notice though that the
second of the above-mentioned patches is only in HEAD, which is why
those same machines are giving scary-looking warnings in the 8.2 branch,
eg wasp on 8.2:

ccache gcc -O2 -Wall -Wmissing-prototypes -Wpointer-arith -Winline -Wdeclaration-after-statement -Wendif-labels -fno-strict-aliasing -g -fpic -I. -I/usr/include/python2.5 -I../../../src/include -D_GNU_SOURCE -I/usr/include/et -c -o plpython.o plpython.c
plpython.c:1996: warning: initialization from incompatible pointer type
plpython.c:1998: warning: 'intargfunc' is deprecated
plpython.c:1998: warning: initialization from incompatible pointer type
plpython.c:1999: warning: 'intargfunc' is deprecated
plpython.c:1999: warning: initialization from incompatible pointer type
plpython.c:2000: warning: 'intintargfunc' is deprecated
plpython.c:2000: warning: initialization from incompatible pointer type
plpython.c:2001: warning: initialization from incompatible pointer type
plpython.c:2002: warning: initialization from incompatible pointer type

Seems we should at least back-patch Marko's fixes into 8.2.

regards, tom lane

Tom Lane wrote:
> I notice though that the
> second of the above-mentioned patches is only in HEAD, which is why
> those same machines are giving scary-looking warnings in the 8.2
> branch, eg wasp on 8.2:

The CVS log message for that says "Allow PL/PythonU to compile on Python
2.5.", but that is evidently not true. Is this just a warning cleanup,
or is there perhaps a 32 vs 64 bit issue that is being fixed there?

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

On 10/15/07, Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:
> Tom Lane wrote:
> > I notice though that the
> > second of the above-mentioned patches is only in HEAD, which is why
> > those same machines are giving scary-looking warnings in the 8.2
> > branch, eg wasp on 8.2:
>
> The CVS log message for that says "Allow PL/PythonU to compile on Python
> 2.5.", but that is evidently not true. Is this just a warning cleanup,
> or is there perhaps a 32 vs 64 bit issue that is being fixed there?

It's a crash fix for 64bit python 2.5.

--
marko

Am Montag, 15. Oktober 2007 schrieb Marko Kreen:
> > The CVS log message for that says "Allow PL/PythonU to compile on Python
> > 2.5.", but that is evidently not true. Is this just a warning cleanup,
> > or is there perhaps a 32 vs 64 bit issue that is being fixed there?
>
> It's a crash fix for 64bit python 2.5.

Then we ought to backpatch that to 8.2.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/