On Tue, 18 Mar 2003, Dan Mills wrote: > On Tuesday, March 18, 2003, at 07:50 PM, Charlie Brady wrote: > > >> * Dots are not required in the filesystem (User requested, hence > >> IMAPdir). ... > As I understand it, users have requested the ability to have mailboxes > in the filesystem that are not prefixed with a dot. IMAPdir attempts > to solve that. Hence, the leading dot is not required (as per the > IMAPdir draft).
Which users have requested it, for what reasons, and is it a reasonable request? In order to be secure, reliable and manageable, software should be as simple as possible (but no simpler). For this reason, anything which complicates the software should be avoided if possible. This definitely complicates the software. I'm unconvinced that there is a valid reason for introducing the complication. > >> * There is a need to disambiguate the same mailbox name with and > >> without a leading dot, then. > > > > I just don't accept that. > > It follows directly from the last point. The IMAPdir draft clearly > accepts mailboxes with and without the leading dot. Hence, the two > must be told apart. You can't quote the IMAPdir draft as authority for this, since it is what is under discussion. Why is it necessary for a maildir named "foo" to be visible to IMAP? [This question assumes that for historical reasons the IMAP folder "foo" is stored in a directory called ".foo".] > >> My comments were about the IMAPdir structure and its intended > >> interpretation, not about policy imposed by binc imap. > > > > But they are the same thing. "IMAPdir" (currently) has no existence > > outside bincimap. And I don't see that it is likely to any time soon. > > It doesn't matter wether third parties implement it right away or not. > Having a fully documented, not to mention sane, storage hierarchy is a > Good Thing(tm). The sanity is a key issue here. -- Charlie
