On Wed, Dec 18, 2002 at 11:47:49PM -0600, Michael Elizabeth Chastain wrote:
My testbed gives me 34 KFAIL's on 34 configurations, just fine.
This is native i686-pc-linux-gnu, gcc v2 and v3, dwarf-2 and stabs+.
I use stock FSF dejagnu 1.4.3. I also built dejagnu from the
sourcware cvs "dejagnu" module and did a test run with that.
That works fine too.
I have appended a gdb.sum file for anyone who is curious.
The KFAIL line looks like this:
KFAIL: gdb.c++/annota2.exp: annotate-quit (PRMS: c++/544)
We can't do anything about the "PRMS:" part, that comes from dejagnu.
I have an objection to the name "c++/544". It is way too easy for
this name to get quoted out of context (the context being that it is
a gdb bug in the gdb database). I think this will cause confusion.
I would like to see "gdb/544" here.
Sure, right *now* while we are discussing the issue, everyone knows
that "c++/544" means a gdb bug in the gdb PR database. Wait six weeks
and then quote some people a gdb.sum report that says "c++/482" in it
and see if anyone jumps to the incorrect conclusion that c++/482 means
a bug in the C++ compiler.
I want the C++ part in there. How about "PRMS: [gdb] c++/544"?
Just gdb/NNN is better - identify the bug database and not the category.
Otherwize everytime someone changes a bug category, they have to go
and update the corresponding testsuite kfail entry (Ulgh!). Also, as
michael-c points out, you can't determine if a c++/NNN is gdb or gcc (I
think it is reasonable for kfails's to refer to the gcc bug database).