On Tue, Apr 25, 2017 at 4:51 PM, Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx> wrote: > We wouldn't be using fnmatch(), but I think it's a probably a good > idea for the tests to support a mode where we have to declare > explicitly whether something should also match under fnmatch or not, > so we document the differences. It was that way for a while (when wildmatch code was still "young") then we removed fnmatch tests because system fnmatch was inconsistent in some corner cases and because wildmatch code did not change much anymore. If we start turning it upside down, yes some sort of checks like that may be a good idea (or at least until new code stablizes again). -- Duy