Re: Re: gtk2?

~
> Some time ago we had a request for an embedded product similar to the iPaq
> and asked several people if GTK+2 would be a choice for it. This is about
> three months ago.
> They told us that GTK+2 would not offer that many benefits and that it
> will be even larger than GTK+1.

Libraries _never_ get smaller (see xlib :)

well ok they don't get smaller until enough people start to wonder why
their libraries are so big (see xlib :)

> IMHO we should stick to GTK+1 for now. It has proven to be stable and
> offers enough for us.

Indeed GTK2 adds 2 new significant dependancies (which may or may not be
Needed) namely ATK (accessibility tooklit) which makes it easy to make GTK
apps accessible (i.e. low-sight users, text-speach interface etc) and
pango (www.pango.org) which is a plugggable text rendition subsystem which
makes support for a variety of previously unsupported scripts (i.e.
devangari?,arabic etc. ) for internationalization (as well as being
extensible to support other script) all based upon UTF8 strings, other
stuff under the hood includes better abstraction in GDK (The windowing
abstraction under GTK+) to allow for easier support for a variety of
windowing systems (win32,beos) as well as some native framebuffer code
(which is where there was some hype revolving embedded/handhelds last
year) there is also a bunch of german guys who have ported GDK to their
accelerated framebuffer architecture (directFB) which is completely
independant from the original FB code.

last time i compiled from cvs, i think the whole dependancy size wasn't
horrendously worse than glib/gtk at the moment (that is with -Os on x86)
as ATK and pango are relatively superficial. although i suspect they might
have a detramental affect on relative runtime sizes.

From an application programmers POV i really can't see a huge reason for
us to switch to GTK2 for a while (like until after it is actual
toolkit/API changes are pretty minor, mainly just consolidation/cleanup.

my opnion is :
it aint finished/stable yet, so don't use it (at least don't encourage
other people to use it), avoids trying to shoot at a moving (and sometimes
stumbling) target. so until gnome 2 comes out i would say it was worth
actually using, esp. considering the fact we may want to apply some
patches to make a HH/smallscreen rendition slightly more paletable.

although saying that it is fairly important that it runs sanely on arm
(from a future point of view) so having somebody testing it on skiff or
something would probably be a good thing. also there is still the issues
of the float code and ditching some of the legacy API code.

I think that was possibly the longest possibly way ever of agreeing with
somebody :)

owen

----
Owen Cliffe, Ph.D. Student, Dept. Computer Science
University of Bath