Hi, I’ve just tried to wade my way through the POSIX pax(1) docs http://pubs.opengroup.org/onlinepubs/9699919799/utilities/pax.html and the pattern matching rules §2.13.{,1.,2.,3.} at http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_13 which isn’t quite fun.
I agree the rules are not applied as-is and mostly undocumented, but the POSIX rules are sometimes very unintuitive and probably prone to break users’ assumptions from ages of use. (Also, note that pax doesn’t support the new “pax” archive format yet, so POSIX compliance isn’t given anyway. But that’s a TODO.) I guess you’re right with that POSIX rules should apply, that is, a period be matched separately, an asterisk (and a range, as ugly as that is) not match a slash. Also, I think §2.13.1. last paragraph applies and backslashes should escape pattern magic characters. I do not agree with your interpretation regarding directories; it doesn’t matter whether the directory matching the pattern is an actual archive member or not when deciding whether to match its whole contents because the pattern matches the directory name. What I’m still concerned is in-the-middle matching, that is, have */foo match not just a/foo but also a/b/c/foo – I wonder whether people would expect that. (On the other hand, POSuX apparently forbids it, so they can go complain there.) I’ll definitively have a look at this eventually, but would still appreciate your input if you so desire. The function in question is pat_rep.c:fn_match() in case you want to have a peek at it, as the codebase changed quite a lot since this bug was initially reported. Thanks for your patience dealing with it. bye, //mirabilos -- 13:37⎜«Natureshadow» Deep inside, I hate mirabilos. I mean, he's a good guy. But he's always right! In every fsckin' situation, he's right. Even with his deeply perverted taste in software and borked ambition towards broken OSes - in the end, he's damn right about it :(! […] works in mksh -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
