Re: [patch] change unresolved to untested in namespace.exp.

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

On Fri, 30 Jul 2010 00:11:47 +0200, Doug Evans wrote:
> If there are no objections, I will check this in in a couple of days.

It is fine with me, it is only for legacy systems with gcc <=4.4 anyway.


> Unresolved means something went wrong with the test run and human attention
> is needed.  This can happen, for example, if a previous test failed.
> But that's not really the case here when the previous test ($test) xfailed.
> Human intervention isn't going to help decide whether $test2 passed or failed.
> 
> ref: http://www.gnu.org/software/dejagnu/manual/x47.html
> grep for UNRESOLVED

UNRESOLVED reasons list also:
 * A test depends on a previous test, which fails.

And the previous test (=GCC version check) failed in this case (generating
XFAIL as a test purely testing the environment IMO should fail by XFAIL).

While UNTESTED lists
 * A test was not run.  This is a place-holder, used when there is no real test
   case yet.
but this testcase is written.

Isn't it more the UNSUPPORTED case?
 * This may mean that a conditional feature of [...] a compiler, is not
   implemented.


> [I don't see why the code doesn't just try $test2 anyway, regardless of
> whether $test failed, but I'm treating that as a separate issue

Because it would PASS despite the code it tries to test is wrong.

# FSF GCC <=4.4 creates unqualified DIE "cOtherFileClassVar" ignoring the
# namespace the same way older GDB did.

That is GCC wrongly generated the symbol without a namespace and GDB wrongly
looks for symbol without a namespace.  Which really works together but
I cannot call it a testcase PASS.


> - I don't know the history of this test.
> But if folks want, I can change the test to do that.]

Its history is very short:
	Re: [patch 1/2] Search typedefs in namespaces also in other files
	http://sourceware.org/ml/gdb-patches/2010-06/msg00663.html


Thanks,
Jan