[RFA] Death to try_named_locale!

Paolo Carlini pcarlini@suse.de
Tue May 17 16:37:00 GMT 2005


Hi,

maybe I'm pretty close to the final goal and, before completing the
boring work on the testcases, want to check with you that I'm not doing
something stupid: as you can imagine, I'm learning a lot, still I'm
largely takinf inspiration from existing DejaGNU code and my
understanding is missing many details.

As you can see, I decided to have something completely inside
libstdc++-v3, because I'm actually using the C++ compiler AND the
library itself (in particular the model-dependent locale code): we have
to both check that the complete set of named locales is available (this
could be done in pure C + <locale.h>) AND that the locale model supports
the use of named locales. Can you see anything wrong with this?

Otherwise, I'm the very same scheme used in
gcc/testsuite/lib/target-supports.exp by check_vmx_hw_available or
check_effective_target_broken_cplxf_arg.

Unfortunately, the interesting bits are at the end of the patch, after
some already converted testcases.

Paolo.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_namedlocale_itw
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20050517/005a3105/attachment.ksh>


More information about the Libstdc++ mailing list