PR 37159

[Dennis Wassel]

Hi list,

I'm having a problem here and would be grateful for some help. See
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37159

I've managed to concoct a patch and testcase (my first, whoohoo!) and
they seem to Do The Right Thing on my Virtualbox 32-bit Ubuntu 8.04.
My troubles are with HAVE_GFC_REAL_16 and {
dg-require-effective-target fortran_large_real }.
If the macro is defined (it is not on my machine), the seed needs to
have 12 elements, otherwise it's 8. This means I need to create two
disjoint testcases, depending on whether HAVE_GFC_REAL_16 is defined
or not. Which testsuite incantations do I need to accomplish that?
fortran_large_real does not seem to be what I think it is, since the
large_real_* tests are executed and PASS, although I cannot select
kind=16 (which makes an unsurprising amount of sense, given that
HAVE... is undefined). I'm quite puzzled...

--Dennis