2011/5/6 Raphael Hertzog <hert...@debian.org>: > On Fri, 06 May 2011, Ondřej Surý wrote: >> >> If you want this fixed, just ping me, it's easy to add something like >> >> "&& (....->name[namelen] == \0 or ...->name[namelen] == '/')", and >> >> I'll fix it at both places. >> > >> > Ah nice catch! I've pushed a fix for this (commit >> > 2c9a342dc4e1ad3e9e58ac89957b9068664d1930). Thanks! >> >> Cool. Also you're right that you don't need to check for \0 in >> dir_is_used_by_pkg, but I am unsure if you can do the same check (/foo >> not matching /foo in other package) is dir_is_used_by_others. In fact I >> think it's wrong... If two packages contains /var/foo/bar/ and both >> depend on this directory to run (f.e. creating files there on the first >> run or only when issuing 'dump' command, etc...), you will remove it on >> first package removal if it's empty thus breaking the second package. > > dir_is_used_by_others does not check the filename in the same way, in fact > it iterates the list of packages associated to the namemode. > > So the problem you mention doesn't exist at all.
That's what happens if you try to use iPhone as full functional device. I checked only the mentioned commit and mixed up dir_has_conffiles and dir_is_used_by_others, because I have remembered that I found it on two places. And you're right and the problem doesn't exist in either. Sorry for not checking harder :). O. -- Ondřej Surý <ond...@sury.org> -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
