Re: mailbox stream

[Alain Magloire]

> 
> 
> There seems to be an inconsistency in handling of mailbox->stream
> member between mailbox.c and mbx_mbox.c, mbx_mboxscan.c. For example,
> if one calls mailbox_set_stream(mbx, stream), the stream is associated
> with the mailbox folder, i.e. mbx->folder->stream, whereas all
> functions in mbx_mbox.c, mbx_mboxscan.c operate on mbx->stream
> directly. This effectively makes mailbox_set_stream() useless.

Yes,

> I suppose the functions in mbx_mbox*.c should obtain stream pointer
> via mailbox_get_stream instead of directly accessing mailbox->stream.
> Am I missing something?

No, It is a confusing scheme that should be settle.  Let me explain the
problem for a remote mailbox, such as IMAP4 or POP3.

Because we need to make a distinction between the stream that represent
the mailbox and the stream that represent the channel or the protocol
(i.e the socket).  So if I do this:

mailbox_get_stream (mbox, &stream);
while (stream_readline (stream, buffer, sizeof buffer, offset, &n) == 0
       n > 0)
 {
    /* write the buffer to a file  ... etc .. */
 }

Meaning I'm expecting to when reading from a stream that I got from
a mailbox that I'm dowloading _The messages_  not receiving things
like "+OK LIST 1 234" or "* SEARCH 1 2 3\ng002 SEARCH completed\n"

To get/set the __tcpip__ stream we should have another function call
I thought of

mailbox_get_carrier (mailbox, stream_t *);
or
mailbox_get_channel (mailbox, stream_t *);


Or we could decide, forget it, this type of functionnality is not need it
they should always read messages via the message_t and
mailbox_{s,g}et_stream () means the to give the __tcpip__ stream.

I'll gladly cleanup the code once we know which path to take.

Any suggestions?


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