Hi,

would it be possible to add .asm to the extensions? I attach a diff that
I have tested with VS Express.

Distutils is frozen, switching to distutils2

Could you add tests? There are helpers in distutils2.tests.support than deal with temporary files, and we use unittest2 which provides decorators to skip the tests on non-Windows.

To me distutils does not support assembler files and for other compilers this is required to build ctypes - see also issue 2942.

> To me distutils does not support assembler files
Do you think it should?

> and for other compilers this is required to build ctypes
Sorry, I don’t understand. Can you rephrase?

Please see my comments to issue 2942 - so I think compilers should support assembler suffixes.

I have a new patch with tests, but I'm not quite satisfied with it.

The remaining problem is that the choice of ml64 or ml is fragile
if a user has executed `vcvarsall xyz`, and we attempt to use the
resulting environment.

For example, if `vcvarsall amd64` has been executed once in a command
line window, a subsequent `vcvarsall x86` will not reset the environment
sufficiently, so ml64 is detected and the x86 build will fail.


But in general this patch should work.

For Visual Studio Express this issue depends on issue 7511.

Minor cleanups in vcasm2.patch.

Further cleanup (sorry for the mail volume!).

(distutils2 bugs have all versions set)

I got a bug report from a user that they encountered this error:

http://tahoe-lafs.org/pipermail/tahoe-dev/2011-April/006312.html

Then a follow-up in which they say they applied the patch from http://bugs.python.org/issue8597 (this ticket is the superceder of that one) and got a different error:

http://tahoe-lafs.org/pipermail/tahoe-dev/2011-April/006316.html

The poster should get the same error messages from:

ml64 /c /Cx x64dll.asm


Perhaps the assembly isn't MASM?

Thanks, Stefan Krah. I posted your comment to the tahoe-dev mailing list:

http://tahoe-lafs.org/pipermail/tahoe-dev/2011-May/006336.html

Also Samuel Neves has posted on that thread.

Steve, this issue has been open for a while:  Do you have an
opinion on whether this is useful to have?

I originally opened the issue because I am using masm in cdecimal,
but there are workarounds.

I'm not against adding the support (to 3.6), though I'd be interested in what the workarounds are and whether it'd be more useful as an example (e.g. "here's how to support a special build step" vs. "all special build steps need to be merged into core").

There are also updates required since we now use distutils/_msvccompiler.py. msvc9compiler.py is only still there so that people who import it directly don't break.

The workaround is telling users to run e.g. ``vcvarsall x64'' before
the build.  setup.py then contains

  extra_objects = ['vcdiv64.obj']
  os.system("ml64 /c /Cx vcdiv64.asm"),

which is run before creating the extension. The extension itself
contains 'extra_objects':

ext = Extension (
        ...
        extra_objects=extra_objects
      )


That has always worked for me.  It isn't a big deal either, since
most people on Windows use installer packages anyway.


I have no strong opinion either way. If you think this adds too much
complexity, feel free to close the issue.

Distutils is now deprecated (see PEP 632) and all tagged issues are being closed. From now until removal, only release blocking issues will be considered for distutils.

If this issue does not relate to distutils, please remove the component and reopen it. If you believe it still requires a fix, most likely the issue should be re-reported at https://github.com/pypa/setuptools