Re: Munging mailboxes

[Alain Magloire]

> 
> Quoting Sam Roberts <address@hidden>, who wrote:
> > In this case there is a message_attachment_filename() that gets the
> > filename of the attachment, but it's not publically exposed. It, or
> > something like it, should be. message_save_attachment() has stub code
> > to use this name if you pass NULL for a filename, but there is what
> > looks like a bug preventing it being used this way. You could remove
> > the check for the filename arg being NULL and see how it works for you.
> 
> I fixed this, see the ChangeLog.
> 

(0)
Cool.
And the code was filter/"sanityse" through GNU indent 8-).

I'm sure Dave's heart will miss a beat, when he see this 8).
Do not know if you noticed but mailutils/mailbox/mime.c still carries
the "evil/alien" indentation ...  Dave, resistance is futile, see the light my 
son.

(1)
 /*FIXME: What is the 'data' argument for? */
  message_save_attachment (part, NULL, NULL);

I think, it was to carry states, if EAGAIN was encounter.  That was probably
the only way to make this async safe.

(2)
Note also, you may want to copy the example/mimetest.c on readmsg/readmsg.c.
I did not have a chance to finish it, the problem was to come up with the type 
of
parsing specify on the command line.  Sergey's proposition
was to reuse the same code in mail/msglist.c:

# readmsg 2[2]     // Print subpart 2 of message 2
# readmsg 1[1-3]   // Print subpart 1,2,3  of message 1

And also to be compatible with ELM readmsg(1)  wich is another parser
for the command line.

# readmsg --parse-style=elm 1 2 3 0  //Pring msg 1 2 3 and the last


Propositions are welcome.