Andreas Aardal Hanssen <[EMAIL PROTECTED]> writes: >On Thu, 30 Sep 2004, Mark Crispin wrote: >>On Thu, 30 Sep 2004, Alexey Melnikov wrote: >>> The mailbox "/m/aaa" doesn't match mailbox pattern "/m/aaa/%", so it >>> shouldn't be returned. >>However, hierarchial name /m/aaa/ would match and could be returned as a >>\NoSelect name. > >Let's make that a "must". Better to keep statements like this consistent.
What if /m/aaa doesn't exist at all? Without the \Nonexistent flag from LISTEXT, a response for /m/aaa/ would be a lie. If nothing else, the client would rightfully expect that if it then created "/m/aaa/bar" the new folder would inherit its ACL from "/m/aaa/", which isn't true. (While I understand the usefulness of \Noselect, Hmm, the relevant sentence in RFC 3501 is Philip Guenther