Re: 2.6 digressions [was Re: release notes pimping [and some random thoughts on 2.6]]

[Havoc Pennington <hp redhat com>]

On Sat, Aug 02, 2003 at 11:26:45PM -0400, Luis Villa wrote: 
> That's where Medusa comes in, of course. ;) [Something else I forgot to
> put on my personal list of 'things that could be really, really cool in
> 2.6'.]

Ah indeed; especially in combination with the "notification when a
file has 0 open descriptors" Alan was talking about, so you can
efficiently keep the index up-to-date.
 
> >  - preferred applications spec that works cross-desktop
> >    (easier if we have cross-desktop IPC system like dbus)
> 
> Possibly a dumb question, but if mime stuff works correctly, why do we
> need the concept of 'preferred applications' at all?

Right, I forgot people on xdg-list are talking about using the MIME
spec for this. I think there's a slight issue with e.g. mapping
"mailto:"; to a MIME type but that can be hacked in to the spec somehow
(I didn't really follow the whole thread).

> >  - unify notification icons and applets into a single cross-desktop 
> >    embeddable object spec; and allow embedding them in desktop
> >    background etc., not just panel
> 
> That's really big picture and will require touching a lot of things. I
> was hoping to focus my list on things that were more within reach :) 

I don't think it's that hard, honestly - at least the embedded object
spec. You don't have to (or want to) break the existing stuff; you
just have to add this alongside.  XEMBED does all the hard stuff,
pretty much. You have to add some interfaces (via some IPC mechanism,
X or otherwise) for the size negotiation and right-click menus and
probably a couple other things, add some sort of
persistence/activation, and there you have it.

By "not that hard" I mean "it'd take somebody good a couple months,
but it would not take them half a year and would not require breaking
a bunch of stuff"

The "make desktop background and panel the same kind of container" is
harder to do completely and would tend to break more things.

> >  - sanitize prefs menu again, consider adding prefs shell 
> >    like XD2 one or something
> 
> We could definitely handle this for the 2.6 timeframe. That said, I'm
> not sure it's really in the category of 'solid code' that you mentioned
> earlier. ;) 

Yeah that may need a rewrite ;-)

> >  - Cairo (was Xr) vector graphics (needs someone to write printing 
> >    backend; then we can do really cool stuff).
> 
> Isn't this more properly handled at the GTK level?

On the GDK level we were just going to add a dependency on Cairo and
add gdk_drawable_get_cairo_surface() [1] - Owen said he wasn't
planning to do a GDK wrapper API. So there's not much to do there. All
the work is on Cairo itself, then in stuff that uses it such as apps,
canvas, etc.

Havoc

[1] I say cairo_surface() but apparently Carl has decided to rename to 
cairo_papyrus() ...