Re: New implementation of EMsgPort

On Tue, 2006-10-17 at 07:19 -0400, Matthew Barnes wrote:
> On Tue, 2006-10-17 at 09:59 +0200, Philip Van Hoof wrote: 
> > This implementation has been merged with the camel-lite which is going
> > to be used by tinymail. This adds another tester for your patch.
> 
> Excellent!  Let me also direct your attention to bug #359979, which
> aims to simplify the EMsgPort implementation even further.
> 
> http://bugzilla.gnome.org/show_bug.cgi?id=359979

All this stuff is very interesting Matthew. Kinda looks a lot like what
I had in mind myself in terms of changes that I'm planning to make
sooner or later.

I will try to merge as much as possible with the version that I will
link tinymail with.

I'm at this moment negotiating with some people how the Camel will be
done (packaged and/or shipped) in tinymail. It will most certainly use a
Camel with a lot changes that likely aren't going to be available in
upstream Camel when I will do my release.

I'm for example definitely planning to use the mmap patches that I did a
few weeks ago (about the folder-summaries in Camel): 
	
Intense measurements (by creating small kernel modules, I know that mmap
isn't easy to correctly measure) have shown me that they do make a very
significant difference in memory usage.

I'm interested in changes like yours. Not because I want to make things
more unstable (less tested) for tinymail ;), but because I believe some
inner core parts of Camel have design problems and others are
replaceable with excellent existing implementations in glib (like
EMsgPort <-> GAsyncQueue, EMutex <-> GMutex, GThread <->  EThread and
finally of course CamelObject <-> GObject).

Feel free to keep me updated by adding me to the CC of new such bug
items or post about it on this mailing list.


-- 
Philip Van Hoof, software developer
home: me at pvanhoof dot be
gnome: pvanhoof at gnome dot org
work: vanhoof at x-tend dot be
blog: http://pvanhoof.be/blog