Re: What is the most suitable way to exclude test cases for specific targets

["Björn Haase" <Bjoern dot M dot Haase at web dot de>]

Mike Stump <mrs@apple.com> schrieb am 02.01.05 18:49:59:
> 
> On Sunday, January 2, 2005, at 06:35  AM, Björn Haase wrote:
> > I have had a look at the test case sources so far in order to find 
> > templates for dg magic comments.  Unfortunately, I did not find 
> > anything suitable so far.
> 
> Then you didn't do the single most useful command:
> 
> 	egrep 'alpha|sparc|frv|x86|mips|elf|linux' `find . -type f`
> 
> ?  :-)
> 
> I am sure after doing that, that you'll find at least three hints.  
> Bear in mind, each part of the testsuite can use a completely different 
> method.
> 

I did of course something similar. I only have looked rather for embedded 
target acronyms like arm or 6811. When you are looking at,  e.g. , 
gcc.c-torture/compile i did not find any method for skipping a test except
for the .x files that seem to be deprecated. The other possibility that I
have seen in other directories is a dg pattern 
            /*  { dg-xfail-if "msg" {"target-triple-with-wildcards"} {""}} */
This one, however seems to declare the test case to be 
of type XFAIL instead of UNSUPPORTED and I have the impression that it
works only in some of the directories. I also did not find out so far what
the second parameter in the curly brackets is good for.

Background of my question is my whish to avoid, that I am preparing a
quick and dirty hack that does solve my present problems but that is
not consistent with the philopsophy that the testsuite maintainers have 
in mind for future additions to the suite. I especially think about the 
question whether one should continue adding explicit expect code.

Regards,

Björn
__________________________________________________________
Mit WEB.DE FreePhone mit hoechster Qualitaet ab 0 Ct./Min.
weltweit telefonieren! http://freephone.web.de/?mc=021201