Re: [Evolution-hackers] evolution-data-server and Migration and GConf Schemas

[Chris Toshok <toshok ximian com>]

I agree, the migration stuff should definitely move into e-d-s. 
Unfortunately, though, at that point we'll have knowledge of the
group/source hierarchy split across evo (for the mailer) and e-d-s (for
addressbook and calendar).  We already have this to some extent now with
libebook knowing the gconf key for the source list, but it's something
to think about.  Is it a bad thing?

I actually have the schema entry for the source list (for the
addressbook at least) in my local tree here.  It's been an outstanding
change for like 3 weeks now I think.

It's too bad we can't split out the portions of the mailer analogous to
libebook/libedatabook (camel, basically, with a little bit from mail/)
and put them in e-d-s as well.  And then move all migration into e-d-s
*ducks from notzed and fejj's glare*

Chris

On Thu, 2004-02-12 at 14:08, Rodney Dawes wrote:
> As I was working on my webcal: handler, I wanted to test it on another
> machine where I didn't have evolution 1.5 installed. So I proceeded to
> build and install the tree up to e-d-s, since that was all I would have
> thought I needed to add a calendar. However, e-d-s isn't doing anything
> to migrate existing calendar/addressbook data, and there are no schemas
> defined for the sources keys in gconf, so when trying to get the sources
> list to add the calendars to, I ended up getting NULL. This made things
> not work. Since e-d-s is intended to be totally separate, I think the
> best thing to do would be to move the cal/contacts migration into the
> data server, and have it signal the GUI to pop up a dialog to show
> the status, and to add schemas for the sources gconf keys, so that they
> will at least exist with an empty sources list, especially if we are
> going to do 1.0 before evolution 2.0, as people might want to start
> playing with the API and various little toys, like the clock applet
> patch, before evo 2.0 is released, and their data is migrated. It
> doesn't seem like it would be hard to do, the schemas stuff at least,
> will be dead simple. Anyone have any complaints, etc...?
> 
> -- dobey
> 
> 
> _______________________________________________
> evolution-hackers maillist  -  evolution-hackers lists ximian com
> http://lists.ximian.com/mailman/listinfo/evolution-hackers