On Tue, Oct 23, 2007 at 09:13:33PM -0400, Matt Rolf wrote:
> >>>From: Karim Nassar <karim giant-rock.com>
> >
> >>>2) If #1 is "no", what are your
thoughts about adding forum
> >>>discussions?
> >>>I think that a neat feature would be that
you could have a
> >>>discussion
> >>>attached to any story. I would be the one
to implement it, and I
> >>>have
> >>>several years of experience that would lend
itself to a nice
> >>>solution
> >>>for your project.
>
> I should have said in my earlier e-mail, if you have
any ideas about
> how this might be implemented in the Bric code base,
I'd be
> interested to hear them.
I have ideas about this, and in fact have started coding...
I had a brainstorming session with Phillip when we were in
Vancouver, and we spent many hours looking at the pros/cons
of implementions up and down the stack from how chats are
stored in the DB to how one would configure the network.
I'm now at a stage of imagining worst case use scenarios and
planning for stability/performance.
There are 3 ways (roughly) that I can think of approaching
this:
1) Complete integration, where this is deeply-tied into
Bric... I think this actually _not_ the route to go at all,
as it's blurring of the data-mangling that Bric does behind
the scenes, and front-end production _serving_, which Bric
doesn't directly deal with.
2) "Wrapping" a solution around Bric output --
this is what Allan did when he wrapped the Bric output in
Drupal processing. The problem with this is
a) performance
b) dealing with multiple paradigms dealing with data
3) Co-processing -- (the route in using) is something that
sits "along side" the production server, and does
it's processing, feeding back to Bric. In this case
Bricolage is still quite seperate from the actual serving of
content (as it should be, in my opinion), but the results of
the comments/forums are fed back to Bric for integration on
next publish. In my imagination, this allows
a) High performance
b) Better seperation of the front-end/back-end processing
c) Even better integration than a "wrapper"
method
> I agree that it would be interesting, I'm just not sure
about the
> feasibility.
I'll keep everybody posted as events warrant...
> -Matt
--
-bch
http://www.methodlogic.net
|