On 11 Jun 2011, at 18:47, Nikodemus Siivola wrote:
On 11 June 2011 19:41, Pascal Costanza pc@p-cos.net wrote:
Just the one in faslcache. If I ask for that particular file directly (foo:bar;quux.fasl), it will only find the first one and cannot find the second one, so directory returning the second one is misleading.
I have to say that I don't particularly disagree -- but that would be have been a more radical change than just fixing the bug in pathname-intersection: something to think about a bit more first.
Just to stress this again: I think directory returning the second name is not only misleading, but also incorrect with regard to what the HyperSpec says. The HyperSpec says that directory "[d]etermines which, if any, files that are present in the file system have names matching pathspec". The second file doesn't have a name that matches the given path.
Directory should do something that is equivalent to translate-logical-pathname, it seems to me, because that is what determines the semantics of logical pathname translations and makes such pathnames unambiguous. (cf. also the non-commutativity of pathname-match-p)
Pascal
-- Pascal Costanza The views expressed in this email are my own, and not those of my employer.