[v2] reshuffle testsuite/22_locale

[Benjamin Kosnik]

>	Did something get dropped during the libstdc++-v3 testsuite
>reshuffle.  I am seeing 79(!) new failres in directory 22_locale.  The log
>file shows the same message for each:
>
>PASS: 22_locale/time_put/put/char/wrapped_locale.cc (test for excess errors)
>spawn [open ...]
>terminate called after throwing an instance of '__gnu_cxx_test::environment_variable'
>  what():  LC_ALL forja_JP.eucjp
>FAIL: 22_locale/time_put/put/char/wrapped_locale.cc execution test

(Damn, the verbose terminate handler is cool.....)

Right.

There were about 500 new files added. That some of them fail on some
targets is not unexpected, and should not necessarily freak you out.
Most of these failing tests were marked XFAIL before.

The goal with this re-organization was:

1) one test per file, more or less

2) try to get "C" locale tests for all facets on all targets active

3) skip wchar_t tests if the library isn't configured for wchar_t: in
the past, these were silently skipped, but marked as passed.

4) move named locale tests into separate files

Now, for you:

terminate called after throwing an instance of '__gnu_cxx_test::environment_variable' what():  LC_ALL for ja_JP.eucjp

Means that testsuite/testsuite_hooks.cc: 

 void
  run_tests_wrapped_locale(const char* name, const func_callback& l)
  {
    ...
  const char* res = setlocale(LC_ALL, name);
    if (res != NULL)
	{ 
         // do test
        }
    else
     throw environment_variable(string("LC_ALL for") + string(name));
}

Which means that the attempt to set LC_ALL to "ja_JP.eucjp" failed.
Perhaps this locale doesn't exist? Who knows.

I think the log message is pretty good, actually, and better than just failing.

I'm not sure what to do in this case. One of the things you could try is
just removing the throw of environment_variable, above, to remove noise
in the testsuites. Another would be to edit the list of testsuite files
to not attempt named locale tests. 

>	Also, I see the same c99_classification_macros_c.cc failure as
>John sees on HP/UX.

Whoops. This needs to be xfailed on linux, but I guess this passes on
AIX and HPUX. Perhaps you could figure out how to do this? I have no idea.

-benjamin