[EVALUATION] - E02 - Support for MinGW Open Source Compiler

[Tim Peters]

[Tim Peters]
>> Well, I'm a Director of the Python Software Foundation, and my view is
>> "the more platforms the merrier".

[Ilias Lazaridis]
> I extract: "you are intrested, that the source-code-base compiles
> directly with MinGW (and other compilers)".

Sure, I'm in favor of that.  I'm also in favor of world peace, for
that matter <wink>.

> Thus you should be intrested, that existent patches are incorporated
> into the source-code-base.

That one doesn't follow.  It follows that I'd like to see existing
patches _reviewed_, but not necessarily that I'd be in favor of
incorporating them if I had time to review them myself.

> The suggested process ist: use of #defines whenever possible, to avoid
> influence on the existent behaviour of the code.

Patches are reviewed on technical merit, balancing the tradeoffs;
#defines are actually discouraged when it's possible to do a thing
without introducing platform-specific #ifdefs.  A problem is that a
patch won't get reviewed unless a volunteer does a review, and we've
got an increasing backlog of unreviewed patches because of that.  The
most effective way for a person P to get their patch reviewed now is
for P to volunteer to review 5 other patches first.  There are a few
Python developers who have promised, in return, to review P's patch
then.

>> But I'm not paid to work on Python, and I don't have time to
volunteer to help
>> MinGW along, so I don't anticipate that I'll do anything here beyond writing
>> this reply.
 
> You have done already very much.
>
> But should should take some time to evaluate community needs.

I don't know what that means, but plausible meanings sound futile.  No
matter what I think "the community" needs, it's not going to happen
unless somebody else does the work:  I can't tell anyone else what to
do.  Heck, I don't even want to.

It sounds like you might want development driven by some kind of
marketing study.  Nothing wrong with that, if so, but it's not how
open source works.  An entity like the Python Business Forum would
presumably be more open to that tack (although I doubt the PBF would
have a natural interest in MinGW).

"Minority platforms" generally don't get far unless a truly dedicated
volunteer shows up.  For example, Jason Tishler does an excellent job
on Python's Cygwin port, as does Andrew MacIntyre on OS/2 EMX, and
they've both done so for years.  Nobody asked them to do this (AFAIK),
it's more that nobody could _stop_ them from doing it. They're
motivated by love of the platforms they take care of.  In the absence
of anyone willing to pay someone else here, that's what's truly
needed.

> ...
> This effort could be driven by the intrested community members (which
> obviously exist).

Then maybe they need to be better organized, and/or more assertive in
pushing their interests.  If someone is getting left behind here, they
should speak up on the python-dev list.

...
> Now, can you please tell me the process I have to follow to suggest the
> following (to the PSF or to the programmers or to the decision takers),
> possibly to get at least a vote on it:

No such thing will happen -- forget that.  For MinGW to be supported
forever, it's necessary and sufficient that a specific person
volunteer to support MinGW forever.  If that person goes away, so does
the support they provided; it's the same story for Cygwin, and even
for Linux and native Windows.  A difference is that Linux and native
Windows attract more than enough volunteers so that ongoing support
seems statistically certain.  But, e.g., if Andrew MacIntyre went
away, I wouldn't bet on OS/2 EMX support continuing.

> "Please ensure that the source-code-base compliles directly with MinGW.
> The suggested process is to:
> 
>   * provide the infrastructure
>     (e.g. mailinglist, issue- tracking-category,... )
> 
>   * Notify the community about this subproject to channelise efforts
> 
>   * include existing MinGW specific patches
> 
>   * ensure future verificatioin of changes,
>     * optimal:    due to an automated build-system
>     * or simpler: due to community-feedback
> "

If a specific person or group wants to volunteer to do all that, year
after year, they can start doing it today.  The PSF won't do any of it
(although the PSF will fund and arrange to run the Python website, and
one way or another supply a bug tracker, source-control system, and
other infrastructure for keeping the Python project as a whole
running).