Python 2.4 killing commercial Windows Python development ?

[A.B., Khalid]

Tony Meyer wrote:
> [problems with dependency on msvr71.dll snipped]
>
> One option is to create a Windows Python 2.4 installer that includes
a
> Python not built with MSVC7 - for example gcc or MSVC6 - which
doesn't have
> the dependency on mscvr71.dll.  Both VC6 and gcc are feasible,
although
> there may be a reasonable amount of work required.
>
> If the installer proved popular, then I'm sure the python-dev people
could
> be convinced (for 2.5, perhaps) that there should be an official
release of
> this type, much like there's a separate 64 bit installer available.
> (Assuming that patches could be submitted that made the build process
as
> seamless as the existing one).
>
> This seems a much more achievable goal than anything involving a
meeting
> with Microsoft...
>
> =Tony.Meyer


Kindly note that the Python source distribution does include project
files for building Python 2.4 with MSVC6. Add to that the fact that
with pyMinGW[1] one can build yet another Windows distribution not
dependent on mscvr71.dll and some of the logic about not upgrading to
Python 2.4, IMHO, just goes away.

An official release of installers for either or both versions would I
think complicate matters: more choices translate to more confusion.
Needless to say that extension authors (for Python 2.4) would then need
to make two binaries for every extension they release for Python 2.4:
one for the mscvr71.dll dependent Python distribution, and another one
for the mscvrt.dll dependent version(s). This I think would hurt Python
and its users.

The solution is to have those that know enough to really need to build
Python on their own according to their requirments. They would then
have to deal with compiling the Python 2.4 extensions themselves, of
course. But this would make things simple and hopefully address the
needs of everyone.


Regards,
Khalid



[1] pyMinGW:
http://jove.prohosting.com/iwave/ipython/pyMinGW.html