Re: [gedit-list] The session stuff and libgnome dependies
- From: Paolo Borelli <pborelli katamail com>
- To: Mikael Hermansson <mike tielie gmail com>
- Cc: gedit-list gnome org
- Subject: Re: [gedit-list] The session stuff and libgnome dependies
- Date: Thu, 27 Dec 2007 03:58:10 +0100
Il giorno mer, 26/12/2007 alle 21.22 +0100, Mikael Hermansson ha
> I decided to forward the bugreport
> http://bugzilla.gnome.org/show_bug.cgi?id=500922 to the gedit-list maybe
> this is intesting for those who dont is CC in the bugreport :-)
> How should we replace the rest of libgnome[ui]?
The road ahead to get rid of many of the current dependencies is pretty
clear, it just need someone doing the work... and yes, thanks for your
patch, it is a great starting point! As you saw I already incorporated
some bits of it and I'd like to get it all committed soon.
But apart from the short term compile time option to disable libgnome,
we really need to plan how to kill it for good by default...
Apart from session support (see below) the blocking feature to get rid
of libgnome is authentication support for remote files: this will be
solved switching from gnome-vfs to gio, but as I say below this is a big
task and not something that can be done in an afternoon.
> Here is my thoughts:
> 1. Session managment
> The problem with the session managment has been discussed for years now.
Session management support must be divided in two features:
1) Notification about the session being closed (the one that lets you
save unsaved document etc)
2) Session serialization (the one which restarts gedit with all the
document that were open in the last session etc)
(1) is something that IMHO we cannot absolutely do without, (2) is
something that has been discussed many times... I personally do not use
it, but I know some people really like it.
While I agree that having something in gtk would be very good, I think
that cut&pasting the libegg code is ok in the shorter term, with the
goal of verifying that code and make it ready for gtk... What needs to
be done here is to review the code in libegg and the patch for gedit
that already exists and test the hell out of it.
> 2. Icon stuff:
Here switching to gio will solve the issue, but we really need to switch
the whole program to gio, not just the icon lookup, so it's a fairly big
> 3. Gnome help
> About the gnome_help #ifdefs pbor did'nt want to kill gnome_help until
> we are sure the new code works.
It's not about being sure the code works: as soon as we can get rid of
libgnome for good I am all for having the code to spawn yelp manually, I
just said that until we have still to keep libgnome we may as well keep
this ifdef too and avoid the duplicated code in the packages shipped by
Just to finish the "state of art" overview, I'll say that there are
other bits and pieces that I'd love to see replaced:
- first and foremost, the printing code: we really need to switch to
gtkprint and remove the dependency on gnome-print, but this need some
work especially since we want to keep our embedded preview and we want
to have fully asynchronous code as we have now.
- we can easily switch away from libglade and use gtkbuilder (there is
even a patch in bugzilla), but there are accessibility regressions...
it's really a shame that the code needed to replace libglade is in gtk
but it misses the needed features to actually use it
To summarize the plan for the future is gedit + gtkprint + gtkbuiler -
libgnome -libgnomeprint - libglade, which would mean gedit with the same
featureset as today with just gtk and gtksourceview as dependencies...
it just needs someone doing the work :-)
] [Thread Prev