Created attachment 463929 [details] Simple test case. Description of problem: When using strptime() to match month names, it seems the English names are considered irrespective of locale. Version-Release number of selected component (if applicable): glibc-2.5-49.el5_5.7 How reproducible: Every time Steps to Reproduce: 1. Compile the attatched test case: make cb 2. Run it: ./cb Actual results: Unexpected match: tm.tm_mon = 9 Expected results: No match, ok! Additional info: Combined with the non-greedy matching in current RHEL5 glibc, see bug 657570, this causes problems trying to match November in the Finnish locale. But even if that bug is fixed, there could be locales where this still cause incorrect behaviour. If there is a locale where "ma" is a short name for a month, for example.
This is by design.
Is there some rationale written for that somewhere? It did look like a bug to me. If I was parsing for a date in a particular locale, I would have expected to discover if I got an input which wasn't a date in that locale, but maybe happened to match English. And it is not without risks. One example is as I explained in the "additional info" section.