> Didn't we decide that if there's an unfixed bug, you file a bug report,
> but you don't check in a test that will fail?
I've marked up the tests with commented-out setup_kfail forms that
refer to the bug number, and included the bug number in the test
name. I thought that's how we'd decided to handle these.
OK, your recollection is echoed by Andrew.
I'm happy now. Please check it in.
Hmm, thinking about it, the message proper, until setup_kfail is added,
should include text like ``(known bug gdb/NNNN)'' so people don't go
chasing after known problems.