On Wed, 26 Apr 2006, Daniel Tiefnig wrote:
> Philip Hazel wrote:
> > This bug has occurred because I never considered the possibility of
> > delivering into anything other than the new folder; I hadn't thought
> > of direct delivery into subfolders.
>
> So, do you think we could get an expanded boolean option that tells exim
> it's delivering into a subfolder?
Yes, we could, but it will be very clumsy. I haven't thought of a tidy
way of doing this. Consider the case when the transport ends up with
directory = some string expansion
where the expansion sometimes yields /some/maildir (for a non-folder
delivery) and sometimes yields /some/maildir/.subfolder - depending on
the local part or local part affix, or on $address_data, for example.
Then the new boolean option will have to do exactly the same expansion
plus a messy test in order to decide whether to yield true or false.
> If it evaluates to TRUE, exim should create the "maildirfolder" file if
> that does not exist, and include its parent directory while calculating
> maildir quota usage. (Creation of "../new", "../cur" and "../tmp" may or
> may not be desired.)
Exim already recognizes existing maildirfolder files for quota
calculations. (And there is also the quota_directory option.) The issue
is just about whether to create them.
Hmm. I suppose one approach would be (subject to a simple boolean
yes/no) to automatically create maildirfolder if the last component of
the delivery directory starts with a dot. Would that work, do you think?
--
Philip Hazel University of Cambridge Computing Service
Get the Exim 4 book: http://www.uit.co.uk/exim-book
