Re: [Distutils] Patch for your setuptools issue

Tue, 02 Jun 2009 06:18:31 -0700 

Tarek Ziadé wrote:

On Tue, Jun 2, 2009 at 12:58 PM, Michael Foord
<[email protected]> wrote:

This failure happens because in line 136(ish) of setuptools.command.test you
are calling unittest.main passing in testRunner=None.

I don't think this is caused by changes to unittest itself and would have
failed with the previous versions. Perhaps changes in distutils or
setuptools?


The test pass under 2.5, 2.6, so there must be something different in
the environment


If I run it with Python 2.6 I get the same problem ('python setup.py test').

It works fine under 2.5 for me though and testRunner is still None.
Ah - I see the problem. In Python 2.5 testRunner defaulted to None and main created its own TextTestRunner. In 2.6 the default was changed to be the TextTestRunner class. Unfortunately 2.5 needs either None or an instance of a runner.
The easiest thing to do might be to create a dictionary of keyword arguments and omit testRunner if it is None. Call main with unittest.main(**args). 

Michael




Happy to dig further into it if needed.


Thanks for the help, I guess we need to dig into it deeper on our side

Cheers


Michael



Tarek Ziadé wrote:

On Tue, Jun 2, 2009 at 12:00 AM, P.J. Eby <[email protected]> wrote:


At 09:39 PM 6/1/2009 +0200, Tarek Ziadé wrote:


2009/6/1 P.J. Eby <[email protected]>:


At 07:01 PM 6/1/2009 +0200, Tarek Ziadé wrote:


Antoine,

here's a patch for your issue.


Hm.  Did you send this patch before?  I don't recall seeing it.



No, I have just created it. I can add it to setuptools issue tracker
if you think it's useful


The idea is useful; I think there might be another way to do it now that
you've given me the idea.  I'd prefer to do it in a way that doesn't copy
code, i.e., by having __getattr__ return something that makes the
distutils-level behavior the same as before, but not by copying the old
distutils code.  That having been said, this might be a good patch to go
in
the meantime.


I have pushed it here :  http://bugs.python.org/setuptools/issue76

the tests pass using previous python interpreters,

Although it fails with the current python trunk,
it seems that unittest.main behaves differently. I am cc'ing  Michael
for some insights because I think
this has been changed lately.


Traceback (most recent call last):
 File "setup.py", line 95, in <module>
   scripts = [],
 File "/Volumes/MacDev/svn.python.org/python-trunk/Lib/distutils/core.py",
line 149, in setup
   dist.run_commands()
 File "/Volumes/MacDev/svn.python.org/python-trunk/Lib/distutils/dist.py",
line 926, in run_commands
   self.run_command(cmd)
 File "/Volumes/MacDev/svn.python.org/python-trunk/Lib/distutils/dist.py",
line 945, in run_command
   cmd_obj.run()
 File
"/Volumes/MacDev/svn.python.org/setuptools/setuptools/command/test.py",
line 121, in run
   self.with_project_on_sys_path(self.run_tests)
 File
"/Volumes/MacDev/svn.python.org/setuptools/setuptools/command/test.py",
line 101, in with_project_on_sys_path
   func()
 File
"/Volumes/MacDev/svn.python.org/setuptools/setuptools/command/test.py",
line 134, in run_tests
   unittest.main(
 File "/Volumes/MacDev/svn.python.org/python-trunk/Lib/unittest.py",
line 1662, in __init__
   self.runTests()
 File "/Volumes/MacDev/svn.python.org/python-trunk/Lib/unittest.py",
line 1744, in runTests
   self.result = testRunner.run(self.test)
AttributeError: 'NoneType' object has no attribute 'run'





I am thinking, btw, that it's time to retire the 0.6 stable branch and
begin
only doing new stuff on the trunk, anyway.


+1

Maybe a final 0.6 release could be done, including the svn 1.6 patch
(or a patch to make the last 0.6.x release not suffering from svn
changes anymore
e.g. by adding a plugin behaviour)


--
http://www.ironpythoninaction.com/









--
http://www.ironpythoninaction.com/

_______________________________________________
Distutils-SIG maillist  -  [email protected]
http://mail.python.org/mailman/listinfo/distutils-sig

Reply via email to