I somehow never received this discussion of yours on sage-devel, and
stumbled on it while browsing trac. Time to answer!
> Sometime in May 2010, John said on sage-devel:
> I had noticed that, after bringing a coupld of files up to 100%
> coverage, it was annoying to get this Error message.
>
> There seems to be nothing about TestSuite in the developers guide, so
> it cannot (yet) be a requirement to implement it!
>
> After �trying TestSuite? I found the page in the reference manual
> (under Miscellaneous, in the section Unit testing for Sage objects).
> But if this is intended to be a requirement, or even a desirable
> feature of every Sage class, there needs to be some better instruction
> on what to do.
The coverage script previously requested a loads/dumps test, and I
replaced it by a TestSuite test. My point of view is indeed that every
class implemented in Sage should be tested with a TestSuite run.
Now I totally agree that it is very annoying when working hard to
improve the test coverage for a class to get "blamed" for all the
issues that the TestSuite call uncovers. So in such a situation, I
find totally acceptable to use:
sage: TestSuite(S).run(skip = ["_test_blah", "_test_blih"]) # see trac #10307
where #10307 would be a new ticket requesting to fix the failing test
in a later patch. Sage is better off with a new TestSuite call that:
- documents what does not work yet and feels itchy, which is good:
someone will eventually want to scratch that itch
- is ready to catch other regressions
than without.
John: do you see a good place in the developpers manual where to add
the above comments?
Cheers,
Nicolas
--
Nicolas M. Thi�ry "Isil" <nth...@users.sf.net>
http://Nicolas.Thiery.name/