Re: [Nmh-workers] exciting new stuff for 2.0 (IMAP proposal)

[Joel Reicher]

> >Sequences are a different problem. Since MH has both public and private
> >sequences, I would argue that the only truly "public" sequence for an
> >IMAP folder is one that is stored on the IMAP server. This could be
> >done with a file on the server, since IMAP allows fairly free form
> >file retrieval, or it could be done with flags where the server permits
> >the definition of new flags. I don't think there would be performance
> >issues with either option, but using flags is definitely better for
> >data integrity where the mail might be accessed by different clients.
> 
> When I looked at this before, a large number of IMAP servers didn't support
> flags, or had a bunch of limitations on them.  I think the only sequence
> you can be guaranteed to support is unseen.

If the server doesn't support whatever mechanism the implementation ends
up using, then it should fallback to private sequences, which are stored
in the context. This is what already happens if a command can't write to
a local folder, and again it solves the storage problem because the folder
directory isn't needed for the private sequence.

Cheers,

        - Joel