Re: [testsuite] add gdb.cp/gdb1355.exp

[David Carlton <carlton@kealia.com>]

On Thu, 18 Sep 2003 10:27:12 -0400, Andrew Cagney <ac131313@redhat.com> said:

> I also intend again proposing that existing XFAILS get yanked.

That sounds okay to me.  I just scanned the XFAILS in gdb.cp; there's
a lot of HP-specific stuff that we have no reason to believe is
accurate, and a lot of branches that are XFAILed for everybody without
any comments as to when we expect that output.  I don't see how either
of those situations are helping us at all.  There are a few XFAILS
which have a bug number associated with them that's around 2400 or so:
these are especially urgent, because it won't be all _that_ long
before the bug number might start looking like a GDB bug number.

The one that looked at first blush like it might be reasonable is the
'setup_xfail_format "stabs"' in method.exp; and, indeed it is
triggered with GCC 2.95.3, but not with GCC 3.2 (even when using
stabs).  In fact, no XFAILS are triggered anywhere in gdb.cp with GCC
3.2 (on i686-pc-linux-gnu, GCC 3.2, DWARF 2 or stabs); on GCC 2.95.3,
I see these:

XFAIL: gdb.cp/classes.exp: ptype class vB (FIXME: non-portable virtual
table constructs)
XFAIL: gdb.cp/classes.exp: ptype class vC (FIXME: non-portable virtual
table constructs)
XFAIL: gdb.cp/classes.exp: ptype class vD (FIXME: non-portable virtual
table constructs)
XFAIL: gdb.cp/classes.exp: ptype class vE (FIXME: non-portable virtual
table constructs)
XFAIL: gdb.cp/method.exp: print this in A::bar (missing const)

where the last one is the one I referred to above, and it's not at all
clear from the other 4 whether they really should be XFAILs or KFAILs
(does the FIXME refer to GDB or GCC?).

So, if the state of XFAILs in other subdirectories of testsuite is
like that for gdb.cp, I don't think they're doing us any good.

David Carlton
carlton@kealia.com