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

[Ilias Lazaridis]

I'm a newcomer to python:

[EVALUATION] - E01: The Java Failure - May Python Helps?
http://groups-beta.google.com/group/comp.lang.python/msg/75f0c5c35374f553

-

I've download (as suggested) the python 2.4 installer for windows.

Now I have problems to compile python extension that some packages
depend on.

I use the MinGW open-source compiler.

-

My questions:

a) Why does the Python Foundation not provide additionally a binary
version, compiled with MinGW or another open-source compiler?

b) Why does the Python Foundation not ensure, that the python
source-code is directly compilable with MinGW?

c) Why are the following efforts not _directly_ included in the python
source code base?

http://jove.prohosting.com/iwave/ipython/pyMinGW.html

above link found in this thread:

http://groups-beta.google.com/group/comp.lang.python/msg/c9f0444c467de525

d) Is it really neccessary that I dive into such adventures, to be able
to do the most natural thing like: "developing python extensions with
MinGW"?

http://starship.python.net/crew/kernr/mingw32/Notes.html

e) Is there any official statement available regarding the msvcr71.dll
and other MS licensing issues?

[see several threads "[Python-Dev] Is msvcr71.dll re-redistributable?"]

http://mail.python.org/pipermail/python-dev/2005-February/thread.html

f) Are there any official (Python Foundation) statements / rationales
available, which explain why the MinGW compiler is unsupported, although
parts of the community obviously like to use it?

http://groups-beta.google.com/group/comp.lang.python/msg/dc3474e6c8053336

-

I just want to understand.

Thankfull for any pointer to official documents / statements.

[google is _not_ a fried here. I like to have a stable development
environment, which is supported by the official projects, thus it can
pass quality-assurance without beeing afraid about every next release.]

..

 

[Michael Hoffman]

a) Why does the Python Foundation not provide additionally a binary
version, compiled with MinGW or another open-source compiler?

I use a binary version of Python compiled with an open-source compiler
on Windows that was provided by someone else.

b) Why does the Python Foundation not ensure, that the python
source-code is directly compilable with MinGW?

Why should they? It already runs on Windows with a freely available
compiler.

f) Are there any official (Python Foundation) statements / rationales
available, which explain why the MinGW compiler is unsupported, although
parts of the community obviously like to use it?

Not to my knowledge. But I would guess because supporting it would
obviously be a lot of work and the core developers have other things to
do they consider more important. They are volunteers, you know.

Why don't you solve this problem and produce a patched version of
Python that does what you want.

[google is _not_ a fried here. I like to have a stable development
environment, which is supported by the official projects, thus it can
pass quality-assurance without beeing afraid about every next release.]

Then you have several options:

a) use a supported development environment
b) do the work yourself to support MinGW
c) pay someone else to do the work

But don't act like the volunteers who develop Python owe you a version
of Python that runs out of the box on MinGW. They don't, anymore than you
owe *me* a version of Python that runs out of the box on MinGW.

Now why haven't *you* produced a version of Python that is directly
compileable with MinGW? Time's a-wasting.

 

[Miki Tebeka]

Hello Ilias,

d) Is it really neccessary that I dive into such adventures, to be able
to do the most natural thing like: "developing python extensions with
MinGW"?

Writing a setup.py and running
python setup.py build_ext --compiler=mingw32
works for me *without* any more work. Things can't get much simpler.

Bye.

 

[Fuzzyman]

I'm a newcomer to python:

[EVALUATION] - E01: The Java Failure - May Python Helps?
http://groups-beta.google.com/group/comp.lang.python/msg/75f0c5c35374f553

-

I've download (as suggested) the python 2.4 installer for windows.

Now I have problems to compile python extension that some packages
depend on.

I use the MinGW open-source compiler.

-

My questions:

a) Why does the Python Foundation not provide additionally a binary
version, compiled with MinGW or another open-source compiler?

It's not necessary.

b) Why does the Python Foundation not ensure, that the python
source-code is directly compilable with MinGW?

Are you sure it isn't ?

c) Why are the following efforts not _directly_ included in the python
source code base?

http://jove.prohosting.com/iwave/ipython/pyMinGW.html

above link found in this thread:

http://groups-beta.google.com/group/comp.lang.python/msg/c9f0444c467de525

d) Is it really neccessary that I dive into such adventures, to be able
to do the most natural thing like: "developing python extensions with

MinGW"?

http://starship.python.net/crew/kernr/mingw32/Notes.html

Not very difficult. The mingw compiler *is* supported through
distutils. distutils can straightforwardly be configured to build
extensions with mingw. The relevent lib files need converting, which is
also simple.

I did it for Python 2.3. For Python 2.4 I use the free MS optimimizing
compiler. That does need a bit of hacking into distutils, but gain -
not very difficult.

Regards,

Fuzzy
http://www.voidspace.org.uk/python/index.shtml

e) Is there any official statement available regarding the msvcr71.dll
and other MS licensing issues?

[see several threads "[Python-Dev] Is msvcr71.dll re-redistributable?"]

http://mail.python.org/pipermail/python-dev/2005-February/thread.html

f) Are there any official (Python Foundation) statements / rationales

available, which explain why the MinGW compiler is unsupported, although
parts of the community obviously like to use it?

http://groups-beta.google.com/group/comp.lang.python/msg/dc3474e6c8053336

-

I just want to understand.

Thankfull for any pointer to official documents / statements.

[google is _not_ a fried here. I like to have a stable development
environment, which is supported by the official projects, thus it can

pass quality-assurance without beeing afraid about every next release.]

.

 

[Ilias Lazaridis]

I use a binary version of Python compiled with an open-source
compiler on Windows that was provided by someone else.

Can you please point me (and the readers) to this resource?

Why should they? It already runs on Windows with a freely available
compiler.

Obvious: Courtesy [against the userbase needs]

Obvious: Consistency [same code-base across different compiler]

Not to my knowledge.

[...] - (guess & comments)

thank you.

Why don't you solve this problem and produce a patched version of
Python that does what you want.

I'm not intrested in patching.

I'm intrested in a stable environment, supported by the original
implementors.

I need a solid fundament for my development.

[google is _not_ a fried here. I like to have a stable development
environment, which is supported by the official projects, thus it
can pass quality-assurance without beeing afraid about every next
release.]

Then you have several options:

a) use a supported development environment

Requirement: "full open-source tool-chain".

b) do the work yourself to support MinGW

this would be not neccessary, as others do this work already.

My question (that you've ommited) was: why does the python foundation
not include this efforts?

[REQUOTE]

c) Why are the following efforts not _directly_ included in the
python source code base?

http://jove.prohosting.com/iwave/ipython/pyMinGW.html

above link found in this thread:

http://groups-beta.google.com/group/comp.lang.python/msg/c9f0444c467de525

[/REQUOTE]

c) pay someone else to do the work

But don't act like the volunteers who develop Python owe you a
version of Python that runs out of the box on MinGW. They don't,
anymore than you owe *me* a version of Python that runs out of the
box on MinGW.

I think Python is a serious Open Source System, driven by the Python
Foundation.

Serious Open Source Systems should serve the basic needs of their
community, especially if there are many depending systems.

If it is a programming language, the requirement "using an open-source
toolchain" is a rational and valid one.

The Python Foundation ingores this requirement, this way creating a
chain of neccessary manual uncontrolled actions.

This does not increase my trust in python [e.g. as an exchange for JAVA].

Now why haven't *you* produced a version of Python that is directly
compileable with MinGW? Time's a-wasting.

I have stated already that I am a newcomer to python.

[you should really avoid this tenor. Python is not an open-source
project of a few teenies. It's a serious programming-language, which
could be adopted by e.g. more phone-manufacturers (after Nokia)]

-

The Python Foundation could create an official sub-project to create an
automated build target based on the MinGW toolchain. I am sure that many
community members would be more than happy to contribute.

..

 

[Ilias Lazaridis]

[please check your news-client. For some reason, the tag "[EVALUATION]"
was removed]

-

You answer essentially something like "It's not necessary" cause "with a
little hacking it works".

I've found lots of documents, which describe how to "hack around" to
make it work.

I don't want to do "hacking".

I want to develope large scale applications, and for this I need an
stable official version of the python language, either binary or
produced directly out of official sources, completely with an
open-source tool-chain.

That's the reason for my very specific questions, which you have mostly
ignored.

-

copied from another answer:

"The Python Foundation could create an official sub-project to create an
automated build target based on the MinGW toolchain. I am sure that many
community members would be more than happy to contribute."

..

 

[Fuzzyman]

Ilias Lazaridis wrote:
[snip..]

Why should they? It already runs on Windows with a freely available
compiler.

Obvious: Courtesy [against the userbase needs]

Obvious: Consistency [same code-base across different compiler]

Are you aware that the MSVC compiler they use produces tighter code
than gcc ? [1] *Most* users would rather have a faster python than a
python built with an open source compiler.

Particularly as distutils (read Python) can *easily* be configured to
use mingw to build extensions from source - which seems to be your real
requirement.

Regards,

Fuzzy
http://www.voidspace.org.uk/python/index.shmtl



[snip..]
[1] Not knocking gcc - it's just optimsied for portability rather than
speed. If you want to see *a* benchmark, there is a link to one in my
'upgrading python' article. (In the article section at
http://www.voidspace.org.uk/python/index.shmtl )

 

[Michael Hoffman]

Michael Hoffman wrote:

Can you please point me (and the readers) to this resource?
http://www.cygwin.com/


I'm not intrested in patching.

I'm intrested in a stable environment, supported by the original
implementors.

And the core developers are not interested in doing more than what they
have already done without further help (e.g. from you). Surely you can
"not interested" as you have justified your own inaction through it.

This does not increase my trust in python [e.g. as an exchange for JAVA].

You cannot run all Java programs on an open source compiler, so I guess
it's an imperfect world for you. And to get GCJ to run on MinGW you have to
add a lot of patches.

Now why haven't *you* produced a version of Python that is directly
compileable with MinGW? Time's a-wasting.

I have stated already that I am a newcomer to python.

[you should really avoid this tenor.

And you should avoid yours. Your sense of entitlement is palpable.

> Python is not an open-source project of a few teenies. It's a serious
> programming-language, which could be adopted by e.g. more
> phone-manufacturers (after Nokia)]

The idea that MinGW support would affect that is laughable.

 

[Ilias Lazaridis]

Hello Ilias,


Writing a setup.py and running
python setup.py build_ext --compiler=mingw32
works for me *without* any more work. Things can't get much simpler.

looks really simple.

-

but:

the central problem still exists:

"** For a Python which was built with Cygwin, all should work without
any of these following steps. **"
source:
http://www.python.org/doc/2.2.3/inst/non-ms-compilers.html#SECTION000312000000000000000

-

"the problem is that Python binary distributions for MS Windows do not
include import libraries for popular gcc based tools: cygwin and mingw32"
source: http://www.zope.org/Members/als/tips/win32_mingw_modules

-

the solutions is possibly (copied from another answer):

"The Python Foundation could create an official sub-project to create an
automated build target based on the MinGW toolchain. I am sure that many
community members would be more than happy to contribute."

..

 

[Michael Hoffman]

"The Python Foundation could create an official sub-project to create an
automated build target based on the MinGW toolchain. I am sure that many
community members would be more than happy to contribute."

An "official sub-project" for something like this is not necessary. Identify
what needs to be done and create a patch, and it will be accepted if it is
a good patch.

 

[David Fraser]

I'm a newcomer to python:

[EVALUATION] - E01: The Java Failure - May Python Helps?
http://groups-beta.google.com/group/comp.lang.python/msg/75f0c5c35374f553

-

I've download (as suggested) the python 2.4 installer for windows.

Now I have problems to compile python extension that some packages
depend on.

I use the MinGW open-source compiler.

-

My questions:

a) Why does the Python Foundation not provide additionally a binary
version, compiled with MinGW or another open-source compiler?

b) Why does the Python Foundation not ensure, that the python
source-code is directly compilable with MinGW?

c) Why are the following efforts not _directly_ included in the python
source code base?

http://jove.prohosting.com/iwave/ipython/pyMinGW.html

above link found in this thread:

http://groups-beta.google.com/group/comp.lang.python/msg/c9f0444c467de525

d) Is it really neccessary that I dive into such adventures, to be able
to do the most natural thing like: "developing python extensions with
MinGW"?

http://starship.python.net/crew/kernr/mingw32/Notes.html

e) Is there any official statement available regarding the msvcr71.dll
and other MS licensing issues?

[see several threads "[Python-Dev] Is msvcr71.dll re-redistributable?"]

http://mail.python.org/pipermail/python-dev/2005-February/thread.html

f) Are there any official (Python Foundation) statements / rationales
available, which explain why the MinGW compiler is unsupported, although
parts of the community obviously like to use it?

http://groups-beta.google.com/group/comp.lang.python/msg/dc3474e6c8053336

-

I just want to understand.

Thankfull for any pointer to official documents / statements.

[google is _not_ a fried here. I like to have a stable development
environment, which is supported by the official projects, thus it can
pass quality-assurance without beeing afraid about every next release.]

Just to add to all the other answers:

Don't just complain, submit patches and work at keeping them maintained.
If this is done for a while it may be more of an argument for having
them included

David

 

[Diez B. Roggisch]

[please check your news-client. For some reason, the tag "[EVALUATION]"
was removed]

I want to develope large scale applications, and for this I need an
stable official version of the python language, either binary or
produced directly out of official sources, completely with an
open-source tool-chain.

Where does that requirement come from? If you want to create large scale
apps, the price for a msvc++ compiler shouldn't matter. And: Windows is a
non-free platform at first. If you have to or want to develop on top of it,
be prepared to pay. Its as simple as that. If you want something cheaper -
you'll have to put some effort into it. Or use linux.

Additionally, your point is moot because there is no need for python _core_
developers or the foundation to support every imaginable platform/compiler
combination. Instead this can be done by companies - see activestate. So if
you want it, step up and do it yourself so your work _becomes_ the official
mingw port. Community gratitude would be guaranteed.

 

[Ilias Lazaridis]

Ilias Lazaridis wrote:
[snip..]

b) Why does the Python Foundation not ensure, that the python
source-code is directly compilable with MinGW?

Why should they? It already runs on Windows with a freely available
compiler.

Obvious: Courtesy [against the userbase needs]

Obvious: Consistency [same code-base across different compiler]

Are you aware that the MSVC compiler they use produces tighter code
than gcc ? [1] *Most* users would rather have a faster python than a
python built with an open source compiler.
Particularly as distutils (read Python) can *easily* be configured to
use mingw to build extensions from source - which seems to be your real
requirement.

This is not relevant.

The Python Foundation should just ensure, that the source-code-base is
compilable with MinGW on windows, too.

This is a very natural requirement.

There is really no need to 'fight' against this.

There is just a need to cooperate to achieve this.

[...]

[1] Not knocking gcc - it's just optimsied for portability rather than
speed. If you want to see *a* benchmark, there is a link to one in my
'upgrading python' article. (In the article section at
http://www.voidspace.org.uk/python/index.shmtl )

..

 

[Ilias Lazaridis]

Where does that requirement come from? If you want to create large
scale apps, the price for a msvc++ compiler shouldn't matter. And:
Windows is a non-free platform at first. If you have to or want to
develop on top of it, be prepared to pay. Its as simple as that. If
you want something cheaper - you'll have to put some effort into it.
Or use linux.

I will not go into this 'twisting' games.

the requirement "Use of an open-source tool-chain" is nothing special.

Additionally, your point is moot because there is no need for python
_core_ developers or the foundation to support every imaginable
platform/compiler combination.

MinGW is not "every imaginable platform/compliler".

Instead this can be done by companies - see activestate. So if you
want it, step up and do it yourself so your work _becomes_ the
official mingw port. Community gratitude would be guaranteed.

I'm not intrested in creating an distribution.

I provide an analysis of the situation, context: newcomer, disapointed
from JAVA.

One of my questions is:

[REQUOTE][/REQUOTE]

..

 

[Ilias Lazaridis]

Ilias Lazaridis wrote: [...]
Just to add to all the other answers:

Don't just complain, submit patches and work at keeping them maintained.
If this is done for a while it may be more of an argument for having
them included

I do not "just complain".

I've spend already hours with writing down the questionaire [which you
have successfully ignored].

..

 

[Ilias Lazaridis]

An "official sub-project" for something like this is not necessary.

It is.

I hope the officials from the Python Foundation detect that:

* source code should be _directly_ compilable under major compilers.
* community efforts for such a important functionality should be
channelized and organized

Identify
what needs to be done and create a patch, and it will be accepted if it is
a good patch.
[REQUOTE]

>> c) Why are the following efforts not _directly_ included in the
>> python source code base?
>>
>> http://jove.prohosting.com/iwave/ipython/pyMinGW.html

[/REQUOTE]

..

 

[Robert Kern]

Ilias Lazaridis wrote:
[...]

Just to add to all the other answers:

Don't just complain, submit patches and work at keeping them
maintained. If this is done for a while it may be more of an argument
for having them included

I do not "just complain".

I've spend already hours with writing down the questionaire [which you
have successfully ignored].

Why don't you spend hours writing code and submitting working patches,
instead? That's what I did years ago in the original effort to get mingw
to compile Python extensions (one of the, woefully out-dated, web-pages
you cite is mine. I have now edited it to clarify the situation so
others do not come away from it as you did).

Just writing "questionnaires" *is* actually "just complaining."

The answer to most of your questions is, "Because no one has yet
volunteered their time and effort to get the job done."

If this is important to you, you need to step up yourself and get it
done and not expect other people to volunteer their unpaid time to
satisfy your whims.

The open source Python community is driven by volunteerism, not a sense
of entitlement. If this does not appeal to you, then perhaps the Python
community is not the right one for you.

--
Robert Kern
(e-mail address removed)

"In the fields of hell where the grass grows high
Are the graves of dreams allowed to die."
-- Richard Harter

 

[Ilias Lazaridis]

http://www.cygwin.com/

thank you.

as far as I know, the created executables are bounded to the GPL.

Thus this is not intresting to me.

I'm not intrested in patching.

I'm intrested in a stable environment, supported by the original
implementors.

And the core developers [...]

please let them speak for themselves.

This does not increase my trust in python [e.g. as an exchange for JAVA].

You cannot run all Java programs on an open source compiler, so I guess
it's an imperfect world for you. And to get GCJ to run on MinGW you have to
add a lot of patches.

Python is from its nature open-source.

The requirement "open-source-tool-chain" fits naturally.

Now why haven't *you* produced a version of Python that is directly
compileable with MinGW? Time's a-wasting.

I have stated already that I am a newcomer to python.

[you should really avoid this tenor.

And you should avoid yours. Your sense of entitlement is palpable.

Entitlements result out of reason.

I'm just pointing out.

I've stated simple questions [which are still unanswered]

And I've stated rationales.

Python is not an open-source project of a few teenies. It's a serious
programming-language, which could be adopted by e.g. more
phone-manufacturers (after Nokia)]

The idea that MinGW support would affect that is laughable.

The idea that the Python Foundation cares about user needs would affect
that.

The idea that the Python Foundation manages to serve (out of one
source-code-base) many platforms/compilers with binaries, due to an
automated, community-supported build system.

This would affect that.

..

 

[Michael Hoffman]

[REQUOTE]

Oh, I can play that game too:

>> [REQUOTE]
>> Identify what needs to be done and create a patch, and it will be
>> accepted if it is a good patch.

MinGW patches have been accepted before. Start submitting yours. As
you point out, there is stuff on the web that means you will escape the
bulk of the work. But not all of the work.

Since you are intent on whining rather than doing the work I'm
ignoring this thread now. Good luck.

 

[Ilias Lazaridis]

Ilias Lazaridis wrote:

[...]
I do not "just complain".

I've spend already hours with writing down the questionaire [which you
have successfully ignored].

Why don't you spend hours writing code and submitting working patches,
instead? That's what I did years ago in the original effort to get mingw
to compile Python extensions (one of the, woefully out-dated, web-pages
you cite is mine. I have now edited it to clarify the situation so
others do not come away from it as you did).

"This page is extremely out-of-date. Not much applies anymore. This page
only remains because too many people still point to it. PExports may
also still be useful to some people; I don't know. I do not use this
platform any longer."
http://starship.python.net/crew/kernr/mingw32/Notes.html

very gentle.

up-to-date information is important, especially to avoid confusing
newcomers.

[you see: even if I just complain, one positive change has already happen]

Just writing "questionnaires" *is* actually "just complaining."

of course not.

The answer to most of your questions is, "Because no one has yet
volunteered their time and effort to get the job done."

this answer do not fit in most questions.

please review them again.

If this is important to you, you need to step up yourself and get it
done and not expect other people to volunteer their unpaid time to
satisfy your whims.

The open source Python community is driven by volunteerism, not a sense
of entitlement. If this does not appeal to you, then perhaps the Python
community is not the right one for you.

I ask some questions and suggest some things.

Voluntarlily and without beeing paid.

There are many commercial systems around python.

So please stop this volunteerism-stuff.

-

If you like to help me and other newcomers, please give me simple some
answers on the initial questions.

..