Re: [rfa:testsuite} Overhaul sizeof.exp

[Michael Elizabeth Chastain <mec@shout.net>]

Hmmm, let's start by looking at the alternatives.  Suppose someone writes
a new test that revels an unexpected bug in gdb (hi Andrew).

[0] Stqtus quo: we reject the test.
[1] FAIL it:  we accept the test and let it FAIL. 
[2] XFAIL it: we accept the test and mark it with XFAIL.
[3] KFAIL it: we accept the test and mark it with KFAIL.

I really want that test in the test suite.  If the test is in the test
suite, it gets run regularly in a lot of configurations.  Anyone can
check the status of the test just by reading the latest Sunday Project
errors-warnings-fails report.  Someone who tackles the bug has the saves
time when they start and when they finish.  And so on.

My preference order is [2] > [1] > [3] > [0].  All but [0] are
acceptable to me.

I understand that [1] has a problem because it looks similar to a
regression.  I don't see that in my tables because they distinguish
between "old test that PASSed before and FAILs now" and "new test with
no result before and FAILs now".  But I acknowledge that other people
oppose [1].

[3] has a problem because KFAIL does not exist.  If the senior maintainers
decide to approve [3] then I can add it to my analysis tables right away,
but someone will have to add it to dejagnu.

I agree with Daniel Jacobowitz; I think [2] is the right thing to do.
It's also practical because we can start today.

  @item XFAIL
  @kindex XFAIL
  @cindex expected failure
  @cindex failing test, expected
  A test failed, but it was expected to fail.  This result indicates no
  change in a known bug.  If a test fails because the operating system
  where the test runs lacks some facility required by the test, the
  outcome is @code{UNSUPPORTED} instead.

  @item UNSUPPORTED
  @kindex UNSUPPORTED
  @cindex unsupported test
  @cindex test, unsupported
  A test depends on a conditionally available feature that does not exist
  (in the configured testing environment).  For example, you can use this
  outcome to report on a test case that does not work on a particular
  target because its operating system support does not include a required
  subroutine.
  @end table

Michael C