Re: Testcase for PR 6026

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Eric" == Eric Blake <ebb9@email.byu.edu> writes:

Eric> Do I need to ask for permission before checking in simple test
Eric> cases?  Just in case, I'm asking for approval for this one on
Eric> mainline and branch.

I don't remember whether the question has ever come up.
I think a good rule of thumb is that test cases are ok to go in
provided they are known to really test something.

For a release branch we need to be a little more strict.  I think we
should only put test cases there if we plan to put the fix there too.
I'm assuming Alex will approve your patch for this bug, so I'd say
check in the test in both places.

Note that if you're checking in a test case without a planned fix, you
should also check in a .xfail file.  Our rule has been that we should
see 0 FAILs.  That lets us know if there is a regression.  (In theory
people should work on XFAILs, but that doesn't happen too often --
just like people rarely seem to work on older PRs.)

Thanks,
Tom