suspicious conflict in python ports

2015-05-27 Thread Marc Espie
py-oauth2-1.5.211(),py-paste-script-1.7.5() /usr/local/lib/python2.7/site-packages/tests/__init__.py /usr/local/lib/python2.7/site-packages/tests/__init__.pyc Those files have surprisingly generic names. I wonder if they should get packaged at all ? To ponder for python people.

Re: suspicious conflict in python ports

2015-05-27 Thread Adam Wolk
On Wed, 27 May 2015 16:11:13 +0200 Marc Espie wrote: > py-oauth2-1.5.211(),py-paste-script-1.7.5() > /usr/local/lib/python2.7/site-packages/tests/__init__.py > /usr/local/lib/python2.7/site-packages/tests/__init__.pyc > > > Those files have surprisingly generic names. I wonder if

Re: suspicious conflict in python ports

2015-05-27 Thread Francisco de Borja Lopez Rio
On Wed, May 27, 2015 at 05:09:09PM +0200, Adam Wolk wrote: > On Wed, 27 May 2015 16:11:13 +0200 > Marc Espie wrote: > > > py-oauth2-1.5.211(),py-paste-script-1.7.5() > > /usr/local/lib/python2.7/site-packages/tests/__init__.py > > /usr/local/lib/python2.7/site-packages/tests/__init__.p

Re: suspicious conflict in python ports

2015-05-27 Thread Adam Wolk
On Wed, 27 May 2015 17:09:09 +0200 Adam Wolk wrote: > On Wed, 27 May 2015 16:11:13 +0200 > Marc Espie wrote: > > > py-oauth2-1.5.211(),py-paste-script-1.7.5() > > /usr/local/lib/python2.7/site-packages/tests/__init__.py > > /usr/local/lib/python2.7/site-packages/tests/__init__.pyc >

Re: suspicious conflict in python ports

2015-05-27 Thread Adam Wolk
On Wed, 27 May 2015 17:36:44 +0200 Francisco de Borja Lopez Rio wrote: > On Wed, May 27, 2015 at 05:09:09PM +0200, Adam Wolk wrote: > > On Wed, 27 May 2015 16:11:13 +0200 > > Marc Espie wrote: > > > > > py-oauth2-1.5.211(),py-paste-script-1.7.5() > > > /usr/local/lib/python2.7/site-packa

Re: suspicious conflict in python ports

2015-05-27 Thread Johan Huldtgren
On 2015-05-27 11:36, Adam Wolk wrote: For net/py-oauth2 the __init__ file is empty. FWIW, removing those files will cause 'make test' to fail. That said currently the tests also fail, seems I missed that when I submitted this port. Some googling led me to this: http://lists.opensuse.org/archi

Re: suspicious conflict in python ports

2015-05-27 Thread Adam Wolk
On Wed, 27 May 2015 15:39:17 -0400 Johan Huldtgren wrote: > On 2015-05-27 11:36, Adam Wolk wrote: > > > For net/py-oauth2 the __init__ file is empty. > > FWIW, removing those files will cause 'make test' to fail. That said Possibly because in order to be able to import a library from a 'folder

Re: suspicious conflict in python ports

2015-05-27 Thread Dmitrij D. Czarkoff
Marc Espie wrote: > Those files have surprisingly generic names. I wonder if they should > get packaged at all ? Adam Wolk said: > For net/py-oauth2 the __init__ file is empty. www/py-paste-script does > contain some setup code for the tests. In both cases I don't think that > any packages should

Re: suspicious conflict in python ports

2015-05-27 Thread Daniel Dickman
On Wed, May 27, 2015 at 6:14 PM, Dmitrij D. Czarkoff wrote: > Marc Espie wrote: >> Those files have surprisingly generic names. I wonder if they should >> get packaged at all ? > > Adam Wolk said: >> For net/py-oauth2 the __init__ file is empty. www/py-paste-script does >> contain some setup code

Re: suspicious conflict in python ports

2015-05-27 Thread Daniel Dickman
On Wed, 27 May 2015, Marc Espie wrote: > py-oauth2-1.5.211(),py-paste-script-1.7.5() > /usr/local/lib/python2.7/site-packages/tests/__init__.py > /usr/local/lib/python2.7/site-packages/tests/__init__.pyc > > > Those files have surprisingly generic names. I wonder if they should >

Re: suspicious conflict in python ports

2015-05-28 Thread Dmitrij D. Czarkoff
Daniel Dickman said: > I don't think it makes sense for tests to live directly under > site-packages. Patch below moves the tests to a sub-dir for py-oauth2. > > ok? Diff below makes more sense to me. Adds TEST_DEPENDS and removes tests from site-packages. -- Dmitrij D. Czarkoff Index: Makef

Re: suspicious conflict in python ports

2015-05-28 Thread Johan Huldtgren
hello, On 2015-05-28 04:25, Dmitrij D. Czarkoff wrote: Daniel Dickman said: I don't think it makes sense for tests to live directly under site-packages. Patch below moves the tests to a sub-dir for py-oauth2. ok? Diff below makes more sense to me. Adds TEST_DEPENDS and removes tests from si

Re: suspicious conflict in python ports

2015-05-28 Thread Johan Huldtgren
On 2015-05-27 16:53, Adam Wolk wrote: That's interesting. When I reported tests failing I meant www/py-paste-script. Your port passes tests correctly *without* the patch. It's very possible that with all the things I've been doing these past few days (weeks) that something in my environment got

Re: suspicious conflict in python ports

2015-05-28 Thread Dmitrij D. Czarkoff
Dmitrij D. Czarkoff said: > Daniel Dickman said: > > I don't think it makes sense for tests to live directly under > > site-packages. Patch below moves the tests to a sub-dir for py-oauth2. > > > > ok? > > Diff below makes more sense to me. Adds TEST_DEPENDS and removes tests > from site-packag

Re: suspicious conflict in python ports

2015-05-28 Thread Dmitrij D. Czarkoff
As suggested by Daniel Dickman off-list, I looked at how other projects deal with tests in py-oauth2. The only project that actually distributes tests is NetBSD, and they do it exactly as is done in our ports right now - placing them to site-packages/tests.[1] FreeBSD[2], DragonFly BSD[3], Debian

Re: suspicious conflict in python ports

2015-05-28 Thread Dmitrij D. Czarkoff
Dmitrij D. Czarkoff said: > Provided that upstream ceased touching oauth2 in 2011, I believe that > patching tests away in "setup.py" is the most viable long-term solution. Similar diff for www/py-paste-script, this time from upstream: https://bitbucket.org/ianb/pastescript/commits/814fbc493e7a999

Re: suspicious conflict in python ports

2015-05-29 Thread Marc Espie
On Wed, May 27, 2015 at 07:19:41PM -0400, Daniel Dickman wrote: > On Wed, May 27, 2015 at 6:14 PM, Dmitrij D. Czarkoff > wrote: > > Marc Espie wrote: > >> Those files have surprisingly generic names. I wonder if they should > >> get packaged at all ? > > > > Adam Wolk said: > >> For net/py-oauth