Re: [PATCH] gdb/testsuite: add untested message to skip_debuginfod_tests

[Tom Tromey <tom@tromey.com>]

>>>>> "Enze" == Enze Li via Gdb-patches <gdb-patches@sourceware.org> writes:

Enze> Hmmm...Not many.  Actually, from the skip_debuginfod_tests, I see an
Enze> untested message "cannot find curl".  It provides detailed information
Enze> when running the testcase.  Then I thought other spots in the same
Enze> procedure may benifit from it.

Personally I think these predicates should just return the result.  If
a message would be helpful, they can use 'verbose'.

Enze> If the way of my previous patch is unconventional, how about this?

I think it's kind of overkill.  "verbose" in the proc would work just as
well, and I suppose my feeling (unsupported by fact really) is that
normally an "untested" is just ignored anyway, and if you want to find
out what's happening you would have to dig a bit anyway.

Also I question if "untested" is correct due to what the dejagnu manual
says:

UNTESTED
     A test was not run.  This is a placeholder used when there is no
     real test case yet.

I think "unsupported" is what we ought to be using.

If you want to see how I think it should generally work, look at my
branch "rewrite-require".  This changes "require" to be simpler and then
tests can express their static requirements very simply:

require dwarf2 !isnative mumble bleah

Then the main difficulty is converting all the code to use this.
I made a stab at it there.

Tom