Doug Evans <d...@google.com> writes:
> I can only guess, but I suspect the confusion here is because one needs to
> remember that "dg" is just one way to write a gcc dejagnu test.
Understood, but by now the vast majority of gcc testsuites uses dg
only.
> Not every test uses dg, and when dg was added to the gcc testsuite I
> certainly wasn't going to load dg stuff in ${tool}.exp.
I wonder if there's any harm in doing so. Unless the driver uses
dg-runtest or a variant thereof, the additional procs shouldn't cause
any problems.
> For reference sake,
> I think a simple rule of thumb for ${tool}.exp vs ${tool}-dg.exp is if it's
> dg-specific put it in the latter.
True, but the distinction has become quite blurred over the years.
> But no matter,
> I have no opinion on what one does today.
I'll try what happens if I remove all the explicit load_lib
${tool}-dg.exp from the drivers and do the loading in ${tool}.exp
instead.
Thanks.
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University
_______________________________________________
DejaGnu mailing list
DejaGnu@gnu.org
http://lists.gnu.org/mailman/listinfo/dejagnu
