In Hungarian the official abbreviation of some monthnames are longer than 3 chars (4 or 5). Please see the attached patch. Is it right to have month name abbreviations that are longer than 3 chars in libc? It causes troubles in some applications, e.g. mc goes stupid, and listing within ncftp and "ls -l --time-style=locale" looks a bit silly. However, IMHO it's better if libc provides correct monthnames and the application does some workaround if it cannot handle more than three chars. Furthermore, if there existed a format string to strftime() which told to display %b in no more than 3 chars, it would help applications (e.g. mc). (Please consider this a feature request :-)
Created attachment 96 [details] correct hungarian month name abbrevs
Hungarian is not the only locale with month apprevations longer than 3 chars. The nn_NO locale have one "lau " value in LC_CTIME:abday, while the rest of the strings are thre chars. I'm not sure if it is a good thing. As you mention, a lot of programs expect all week day apprevations to have the same length - 3 chars.
There is nothing wrong with longer abbreviations. The important thing is they are understandable by the audience. The rest are bugs in the applications making wrong assumptions.
I admit that my description is not fully clear, but take a look at the patch, it is in the reverse way than you think. Currently glibc uses 3 letter abbreviations for all the hungarian months, though the correct would be a 4 or 5 letter abbrev for some of them (febr, márc, szept). If there's nothing wrong with them (as you say) then please commit this patch. Thanks! :)
Applied to CVS.