Paul Eggert <[EMAIL PROTECTED]> wrote: > One alternative is that 'configure' could complain if "sort" outputs > anything on stderr given a trivial case. Or the tests involving > "sort" could each test for this incompatibility of CoSort, and ignore > stderr only in that case.
I'd say it's rather a task for configure, not for tests themselves. The only question is how to test that sort being used is actually CoSort. Regards, Sergey _______________________________________________ Bug-tar mailing list Bug-tar@gnu.org http://lists.gnu.org/mailman/listinfo/bug-tar