Re: [Evolution-hackers] calendar EPlugin plans

[JP Rosevear <jpr novell com>]

On Thu, 2004-10-14 at 02:42 -0600, David Trowbridge wrote:
> On Thu, 2004-10-14 at 09:01 +0800, Not Zed wrote:
> > On Wed, 2004-10-13 at 01:34 -0600, David Trowbridge wrote: 
> > > Since I obviously haven't communicated my plans properly to everyone,
> > > here's a short roadmap of what I plan to do.
> > > 
> > > 1. EConfig-ize the calendar properties window
> > > 	- this is mostly done. Pending some cleanup, I should have a patch off
> > > to e-p for review soon.
> > > 
> > > 2. EConfig-size the new calendar window
> > > 	- this is pending UI decision
> > > 
> > > 3. Add some API to allow plugins to register new sources and a calendar
> > > component init hook. The goal of this is to allow plugins to detect the
> > > absence of and register new top-level sources, similar to the existing
> > > migration code. This will make it so new calendar backends can be
> > > distributed completely independent of evolution.
> > I don't understand why you need this at all?
> > 
> > Doesn't the e-source-list manage all of these?  You should just be
> > adding a new source the same way other sources are added, and they'll
> > just 'work', wont they?
> It sounds like what ought to happen is a separation of source type from
> the source groups.

Yes, that is my thought as well.
 
> > Once you add the stuff to the new-calendar thing to select the type of
> > calendar, the rest should just follow. 
> > > 4. (maybe) Add some stuff to allow plugins to do size- and position-
> > > requests and custom drawing for calendar elements. This would probably
> > > only be used for esoteric calendars like weather, where the standard
> > > rendering and interaction doesn't really make too much sense for the
> > > data.
> > This sounds to me to be something more fundamental than should be left
> > to 'plugins'.
> The idea would be to allow a plugin to override the default behavior.

This might be a good long term idea, but I'm not sure this should be
before two things happen:

1) Rodrigo's recurrence branch code is finished and merged to HEAD (this
will alter the e-cal-model)
2) The drawing code is cleaned up, its  a giant 20000 LOC mess right
now.

-JP
-- 
JP Rosevear <jpr novell com>
Novell, Inc.