buildout's tests fail for me
I've made a reasonable number of recipes and there's one thing in buildout that I want to fix. In order to fix it, I need to be able to run the tests, of course. But the tests fail for me. Tens of them. (I've posted about this before). The kind of errors you get: Expected: setuptools.egg - zc.buildout.egg Got: - setuptools.egg-link - zc.buildout.egg As far as I can see, there's one root cause for all of them: setuptools is detected as a development egg on line 346 of buildout.py. There was at least one other developer that was plagued by exact the same problem. He was on linux, I'm on osx. So it is not specifically os-dependent. I've tried installing setuptools (globally) in several ways. Whether I use macport's version or the downloaded setuptools egg: no difference. So: - Is this a known issue? - Is this something I really should fix in my own setup? - Is there a known fix that can be applied to the test setup? I'd really like to contribute :-) Reinout -- Reinout van Rees - reinout@vanrees.org - http://reinout.vanrees.org Software developer at http://www.thehealthagency.com "Military engineers build missiles. Civil engineers build targets"
On Tue, Sep 1, 2009 at 3:42 AM, Reinout van Rees
- Is this something I really should fix in my own setup?
- Is there a known fix that can be applied to the test setup?
Yes. Use a clean Python built from source with nothing in site-packages. To build the buildout checkout, use dev.py, which is a special bootstrap used when working on buildout itself. Jim -- Jim Fulton
On 2009-09-01, Jim Fulton
On Tue, Sep 1, 2009 at 3:42 AM, Reinout van Rees
wrote: - Is this something I really should fix in my own setup?
- Is there a known fix that can be applied to the test setup?
Yes. Use a clean Python built from source with nothing in site-packages.
Ok, that helped partially. But then I got an error like: distutils.errors.DistutilsPlatformError: $MACOSX_DEPLOYMENT_TARGET mismatch: now "10.3" but "10.5" during configure Waargh, that was in python2.4-checking code. So I tried running the buildout with python2.4 instead of the 2.5 I've been using: success. I don't know if the python24/python25 difference is known/intended, but in any case: I can run the tests now.
To build the buildout checkout, use dev.py, which is a special bootstrap used when working on buildout itself.
I haven't seen that dev.py documented. Would a "development instructions for buildout itself" be good to add to the documentation? Reinout -- Reinout van Rees - reinout@vanrees.org - http://reinout.vanrees.org Software developer at http://www.thehealthagency.com "Military engineers build missiles. Civil engineers build targets"
On Tue, Sep 1, 2009 at 9:19 AM, Reinout van Rees
On 2009-09-01, Jim Fulton
wrote: On Tue, Sep 1, 2009 at 3:42 AM, Reinout van Rees
wrote: - Is this something I really should fix in my own setup?
- Is there a known fix that can be applied to the test setup?
Yes. Use a clean Python built from source with nothing in site-packages.
Ok, that helped partially. But then I got an error like:
distutils.errors.DistutilsPlatformError: $MACOSX_DEPLOYMENT_TARGET mismatch: now "10.3" but "10.5" during configure
<shrug> Mac OS X weirdity.
Waargh, that was in python2.4-checking code. So I tried running the buildout with python2.4 instead of the 2.5 I've been using: success.
I run the tests on Mac OS X with Python 2.4, 2.5 and 2.6.
I don't know if the python24/python25 difference is known/intended, but in any case: I can run the tests now.
To build the buildout checkout, use dev.py, which is a special bootstrap used when working on buildout itself.
I haven't seen that dev.py documented. Would a "development instructions for buildout itself" be good to add to the documentation?
I wouldn't make it part of the standard documentation. Perhaps a DEVELOPER.txt in the root directory. Jim -- Jim Fulton
On 1 Sep, 2009, at 16:21, Jim Fulton wrote:
On Tue, Sep 1, 2009 at 9:19 AM, Reinout van Rees
wrote: On 2009-09-01, Jim Fulton
wrote: On Tue, Sep 1, 2009 at 3:42 AM, Reinout van Rees
wrote: - Is this something I really should fix in my own setup?
- Is there a known fix that can be applied to the test setup?
Yes. Use a clean Python built from source with nothing in site- packages.
Ok, that helped partially. But then I got an error like:
distutils.errors.DistutilsPlatformError: $MACOSX_DEPLOYMENT_TARGET mismatch: now "10.3" but "10.5" during configure
<shrug> Mac OS X weirdity.
What kind of Python installation are you using? System python, binary installer from Python.org, ... Which version of Python are you using? Do you have MACOSX_DEPLOYMENT_TARGET set in your shell environment? Ronald
On 2009-09-01, Ronald Oussoren
On 1 Sep, 2009, at 16:21, Jim Fulton wrote:
On Tue, Sep 1, 2009 at 9:19 AM, Reinout van Rees
wrote: distutils.errors.DistutilsPlatformError: $MACOSX_DEPLOYMENT_TARGET mismatch: now "10.3" but "10.5" during configure
<shrug> Mac OS X weirdity.
What kind of Python installation are you using? System python, binary installer from Python.org, ... Which version of Python are you using? Do you have MACOSX_DEPLOYMENT_TARGET set in your shell environment?
Set in my shell? No. And this last attempt where I got the 10.3 != 10.5 error was with a freshly compiled 2.5.4 source release from python.org. Plain configure/make/make install (with a --prefix= option). Plain as plain can be. If you google it, there are quite a lot of reports on this. Well, I've got the tests running with python2.4 at least. Hm. I *could* do another build WITH such an environment variable set... Reinout -- Reinout van Rees - reinout@vanrees.org - http://reinout.vanrees.org Software developer at http://www.thehealthagency.com "Military engineers build missiles. Civil engineers build targets"
On Tue, Sep 1, 2009 at 09:42, Reinout van Rees
He was on linux, I'm on osx. So it is not specifically os-dependent.
No problem here:
http://misc.buildbot.securactive.org/waterfall
--
Sebastien Douche
On 2009-09-01, Sebastien Douche
On Tue, Sep 1, 2009 at 09:42, Reinout van Rees
wrote: He was on linux, I'm on osx. So it is not specifically os-dependent.
No problem here: http://misc.buildbot.securactive.org/waterfall
That's exactly the problem: most people don't have a problem. So it took me quite some time to be actually able to work on buildbot itself as all I heard was "the tests run for me". I've added two of the most helpful hints to a new DEVERLOPERS.txt to make sure they're documented now. Reinout -- Reinout van Rees - reinout@vanrees.org - http://reinout.vanrees.org Software developer at http://www.thehealthagency.com "Military engineers build missiles. Civil engineers build targets"
participants (4)
-
Jim Fulton
-
Reinout van Rees
-
Ronald Oussoren
-
Sebastien Douche