Re: imap4d PERMANENTFLAGS

[Alain Magloire]

> 
> Hi,
> 
> I am trying to make imap4d save the state of permanent flags. Many
> clients (e.g. netscape) do rely on this.

> But current API doesn't seem to support the operation.

Yep, there is no provision for it.

> I would propose to add something like
> 
>         int  (*_updateflags)         __P ((mailbox_t));
> 
> to struct _mailbox. Any suggestions?

Ho! I see, you mean since we only set/save the attributes at expunge
the next time the client comes and it did not send and EXPUNGE on the
last session, the attribute will not be set.

OK.  Rewriting the mailbox to save the flags in "Status:" is a lengthy
operation, it would probably be better to do it in batch on the
reSELECT/EXAMINE or on CLOSE.  Not having the RFC at hand, I'm not sure that
the RFC requires the CLOSE be send before a mailbox be properly updated
when it was selected.  I think that SELECTing a new mailbox
automaticaly CLOSE the old one and select the new one.


For a store format like mbox(which is a flat file) it is probably
better to update "en masse", then doing it individually for each message
But for others, like MH or MAILDIR, a file per message, there is no gain,
so how about this:

message_update_attribute (message_t msg)
{
   mailbox_update_attribute (msg->mailbox, msg); /* per message.  */
} 

and
mailbox_update_attribute_all (mailbox_t);  /* en masse.  */

This is just a suggestion,  the names look rather long though 8-)
But what I'm getting at is the flexiblity to update the flags/attribute
per message for the mailbox that has the capability to do it efficiently.
Unix-mbox mail format(one flat file) is desperetly brain-damage, no need
to hurt.

Let me know, your current thoughts so we can reach an agreement.
I can then do an implementation to the new rewriting of mailbox
and backport it to the current or vice-versa.

-- 
au revoir, alain
----
Aussi haut que l'on soit assis, on n'est toujours assis que sur son cul !!!