Rafael Garcia-Suarez wrote:
That assumes that tests are skipped per file, which is not always the
case (sometimes you want to skip only one test, sometimes even to work
around an OS bug that appears only in one specific version). But
reorganizing platform-dependent tests might be a good idea.
Aye, I would want to improve on the Python solution. But maintaining a
config file something like:
[darwin]
t/pmc/foo.t 3 5-7 9 # platform doesn't support libfoo
t/pmc/bar.t 1 42
...
[MSWin32]
t/pmc/foo.t 32
...
would be far simpler than maintaining what we have now, and also
provides a snapshot view of what's broken on which platforms. And, with
Parrot's architecture, a core "Test::More" (or "Test") module could be
aware of which tests should be skipped and make that information
available within the test file, no matter what language the test file
was written in.
There are other reasons for skips/TODOs that don't make sense in a
global config (such as "skip this test if the previous line returned
"foo"), but those could continue as calls within the test file.
I (mostly on behalf on P5P) would find it interesting to see how that
turns out to be implemented in parrot's test suite, so we can blatanly
copy the good ideas and avoid the pitfalls :)
As we have also done in the other direction, so it works out. :)
Allison
