Re: [gmime-devel] Putting GMime into Bongo

[Jeffrey Stedfast <fejj novell com>]

On Fri, 2008-09-05 at 16:13 +0100, Alex Hudson wrote:
> Jeffrey Stedfast wrote:
> > Haha, back when it was Hula at Novell, I had a talk with those guys at
> > lunch when I was out in Provo and they were interested in looking into
> > GMime, but I don't think they ever got around to it.
> >   
> 
> Hehe, I never knew that :) I suspect you're right, I don't remember 
> anyone looking at it (though I was just a Hula acolyte then, rather than 
> a hacker).

I just remember them looking at me across the table and saying something
along the lines of "I didn't know /anyone/ liked writing parsers..." :)

> 
> > DBMail (a free software imap server) is using GMime 2.2.x right now as
> > far as server software goes. There's also Balsa and Pan which use it on
> > the client-side and from what I've been told, Brazil's main ISP and
> > PennState U. both use it in custom spam-filters which filter all mail
> > that comes through.
> >   
> 
> Cool. If it's good enough for them, it's probably good enough for us. 
> The abort() situation in malloc is probably not a deal-breaker - we need 
> a policy on that anyway, and a global abort() might actually be a 
> relatively graceful way of doing it (we have a lot of little daemons, 
> and one overall manager, so having one throw an abort might be an easy 
> way to signal that the whole system's in trouble and we should stop 
> doing things like accepting mail ;).

that works ;-)

> 
> I've asked another one of our hackers, who previously wrote part of one 
> of the parsers :), to take a look at GMime and see what he thinks. I 
> think my main concern at this point is that we would effectively have a 
> lot of code "duplicated" (glib versus bongoutil/others, iconv versus 
> bongostreamio, etc.) but some of this stuff I'd be looking to get rid of 
> in the longer term anyway, and the pain of having multiple MIME parsers 
> (we actually have two) is probably worse.
> 
> We'd probably also want to look at the functionality of the two code 
> bases and see if there was anything that we have that you don't; I 
> imagine the odds of that are pretty small, but you never know :)

Awesome, if you find anything, let me know. You might want to look over
the development releases which have changed the API a bit (2.3.8 is the
latest, I haven't made any adjustments to svn since I released it).

I've been trying to make the overall API cleaner which has resulted in
some functions which used to return a linked-list pointer being changed
to instead return an object which has an API for clearing,
adding/inserting/removing/etc items from it directly, so that I didn't
have to duplicate functionality in the container object.

This is most noticeable with g_mime_message_get_recipients().

I posted a bit about some of the changes I've been making to the
mailing-list (the mailing-list has only been around for a few weeks, so
there's fairly little backlog to read over).

If you guys have any suggestions or features that you need, I'll try to
get them in for the next stable release (2.4.0?).

Jeff