glad to generate some discussions ;-) 2010/12/15 Nicolas Chauvat <[email protected]>
> On Wed, Dec 15, 2010 at 11:09:13AM +0100, Julien Jehannet wrote: > > I'm quite sure we don't want advertise too much this "dark corner" of > > logilab.common ;-) The actual trend is to replace testlib and pytest by > > recent changes from unittest/unittest2 (even if not already in stdlib). > > > > So please, conѕider using unittest2 in favor of testlib for your own > projects. > > If you want to use some "feature" of testlib, please use a copy of the > module. > > What is the dark corner you mention ? > > The plan for logilab.common.testlib is: > > 1/ base it on unittest2 instead of unittest > > which will remove differences like assertRaises (the one I was looking for last Friday Nicolas) and others if any. As long as I am concerned, most of my tests run transparently with both unittest2 and testlib. > 2/ remove/refactor everything that duplicates stuff from unittest2 > > 3/ if there are things left, try to implement them as plug-ins on top > of unittest2 and offer to include them in the next version > > that's what I found interesting in testlib: additional features compared to std unittest and I am sure they are some even with unittest2. One that I would really like is assertions for numpy arrays or at least an ListAlmostEqual; handy when working with numerical data. > 4/ as far as possible, do it transparently so that things that work > with logilab.common.testlib continue working (but maybe printing > out deprecation warnings) > -- > Nicolas Chauvat > > logilab.fr - services en informatique scientifique et gestion de > connaissances > -- Jean-Philippe
_______________________________________________ Python-Projects mailing list [email protected] http://lists.logilab.org/mailman/listinfo/python-projects
