Mod_Python vs mod_wsgi

[ray]

After looking into the most current version of mod_python, I found
that there is little to no support for it - development seems to have
stopped about 1 1/2 years ago.

The last active developer has suggested that users move to mod_wsgi.
Reference:
http://code.google.com/p/modwsgi/

The message can be found at:
http://modpython.org/pipermail/mod_python/2009-July/026279.html

I would appreciate comments.

Ray

[Graham Dumpleton]

Development on mod_python effectively stopped 2 1/2 years ago, not 1
1/2.. Quoting time of last release from mod_python site:

Feb-14-2007 3.3.1 is out and available on the download page.

There may have been a few commits in repository since then, but only
because I got sick of answering questions as to why last release
wouldn't build on latest APR versions and MacOS X. I wouldn't construe
those commits as being an indication of active development. :-)

Graham

[Greg Troxel]

I would appreciate comments.

I am running trac on about 5 boxes, NetBSD 4 or 5, i386 or amd64, apache
2.2.x. Most are running mod_wsgi and perhaps one or two mod_python. My
plan is to make them all run mod_wsgi. I suspect that mod_python leaks
memory into apache, even the master process, but I can't quite prove
that. mod_wsgi is slightly tricky to set up in that you need to install
a small script but once you get it working it is entirely reliable, so
this is in my view not a worry.

So I recommend that you use mod_wsgi.
Graham presumably would have said that too, but I'm disinterested :-)

[Graham Dumpleton]

On Jul 27, 9:39 pm, Greg Troxel <g...@ir.bbn.com> wrote:
> After looking into the most current version of mod_python, I found
>   that there is little to no support for it - development seems to have
>   stopped about 1 1/2 years ago.
>
>   The last active developer has suggested that users move to mod_wsgi.
>   Reference:
>  http://code.google.com/p/modwsgi/
>
>   The message can be found at:
>  http://modpython.org/pipermail/mod_python/2009-July/026279.html
>
>   I would appreciate comments.
>
> I am running trac on about 5 boxes, NetBSD 4 or 5, i386 or amd64, apache
> 2.2.x.  Most are running mod_wsgi and perhaps one or two mod_python.  My
> plan is to make them all run mod_wsgi.  I suspect that mod_python leaks
> memory into apache, even the master process, but I can't quite prove
> that.

Yes, mod_python has a problem in it whereby it doesn't destroy the
interpreter properly in Apache parent process resulting in all the
memory in use by the interpreter being leaked. See:

https://issues.apache.org/jira/browse/MODPYTHON-235

In mod_wsgi, it does destroy the interpreter properly and so doesn't
suffer the same problem.

That said, the Python interpreter itself, even when destroyed, does
leak a small amount of memory depending on which version of Python you
are using.

End result is that for mod_wsgi, you still may see Apache parent
process grow in memory usage if you do a lot of restarts or graceful
restart, rather than complete stop/start.

> mod_wsgi is slightly tricky to set up in that you need to install
> a small script but once you get it working it is entirely reliable, so
> this is in my view not a worry.

Although some like that mod_python can just refer to a installed
Python module, that it does this is a PITA. This is because it is
completely contrary to the Apache resource model and as a result the
SCRIPT_NAME value which identifies where an application is mounted is
usually never correct for mod_python. As a result, applications when
using mod_python usually have to provide a way of fudging things by
telling the application manually where it is mounted.

> So I recommend that you use mod_wsgi.
> Graham presumably would have said that too, but I'm disinterested :-)

As much as I would like to see mod_python vanish, I am not that
blatant about telling people that mod_python is a stinking pile of cow
turd and that mod_wsgi is much sweeter smelling. :-)

Graham

[Jeroen Ruigrok van der Werven]

-On [20090728 03:00], Graham Dumpleton (graham.d...@gmail.com) wrote:
>That said, the Python interpreter itself, even when destroyed, does
>leak a small amount of memory depending on which version of Python you
>are using.

I assume you logged something about this on bugs.python.org, Graham?

--
Jeroen Ruigrok van der Werven <asmodai(-at-)in-nomine.org> / asmodai
イェルーン ラウフロック ヴァン デル ウェルヴェン
http://www.in-nomine.org/ | http://www.rangaku.org/ | GPG: 2EAC625B
I dream of gardens in the desert sand...

[Graham Dumpleton]

On Jul 28, 8:35 pm, Jeroen Ruigrok van der Werven <asmo...@in-
nomine.org> wrote:

> -On [20090728 03:00], Graham Dumpleton (graham.dumple...@gmail.com) wrote:
>
> >That said, the Python interpreter itself, even when destroyed, does
> >leak a small amount of memory depending on which version of Python you
> >are using.
>
> I assume you logged something about this on bugs.python.org, Graham?

No. Apparently it is a design decision in later versions that they
leave some stuff to leak. For whatever reason it is just too hard to
clean up everything.

Graham