[Fontforge-devel] Fontforge UI improvements

[Dave C. <da...@la...>]

Hi!

Is anyone interested in taking this on? :-)

---------- Forwarded message ----------
From: MenTaLguY <me...@ry...>
Date: Wed, 12 Oct 2011 18:56:48 -0700
Subject: Fontforge UI improvements (Was: Re: [Inkscape-devel]
versioning scheme, take 2)
To: Dave Crossland <da...@la...>

Hi Dave,

I'd still like to do Fontforge UI work, though I don't really need
the work at present.  Time might also be a problem for me.  If you want
to try it with someone else, I'm not sure how much detailed
documentation would help, short of actually doing it and documenting the
steps required, but the basic idea as I recollect it was something like:

 1. Port the bespoke UI toolkit from Xlib to Gdk

 2. Rework top-level windows so that they're GtkWindows rather than bare
GdkWindows, with an embedded GtkDrawingArea used for everything

 3. Start reimplementing individual UI elements (beginning with the
outermost layout containers in the widget trees and working inward)
using corresponding Gtk widgets; the remainder continue to draw directly
on embedded GtkDrawingAreas

 4. Again, working out-to-in, begin peeling back the wrappers to work
explicitly in terms of Gtk widgets, if you decided you wanted to do
that.

Step #1 by itself (probably plus a bit of extra cross-platform fiddling
in the non-UI code) would let you get something that worked on Windows
without requiring an X server, though it wouldn't look native.  (You'd
still need to address the problem of a new build process to get native
Windows and OS X executables.  That's probably better done by platform
experts.)

But I don't really know whether the incremental approach is best
approach now, or whether it'd be better to just bite the bullet, freeze
the existing interface, and re-implement it in Gtk from scratch -- maybe
in C++ or Python.  The incremental route will require a level of
combined X and Gtk experience (or at least sheer bloody-mindedness)
that's hard to find.  And least you've already nicely split the the
application UI and backend.

Something else to consider: once Ubuntu switches to Wayland, many Linux
desktops may no longer be running an X server by default.  Gdk/Gtk apps
would have a Wayland backend, but raw Xlib apps wouldn't work without
installing and starting X-on-Wayland.  So in some ways that may make the
task more urgent.  I wish I could take it on right now.

-mental