OO & GNOME, summary & an action request

[Seth Nickell <snickell stanford edu>]

Breaking with the "metaphysics", here is my rough (terribly biased)
summary of points salient to the particulars of the issue. All Sun
quotes are taken from the main Sun page concerning GNOME
(http://www.sun.com/software/gnome), mostly from one paragraph. I am
assuming that this page is not B.S. and reflects the actual desires and
intentions of Sun Microsystems concerning GNOME. The action request is
tacked to the bottom since it seemed somehow disjunct placed before my
long-winded rants ;-) I apologize in advance to those who have already
realized I'm just beating a dead horse (I obstantly continue to write in
spite!). This is all my personal opinion, blah blah blah.

1) At least some fair number of (prominent and otherwise) GNOME hackers
feel OO should not be considered GNOME software if it does not make use
of gnome-libs and gtk+ (as they seem appropriate to an office suite) for
widgets and such. 

2) Given that Mozilla, another prominent and extremely useful piece of
desktop software, is not going to be included in GNOME 1.4 for very
similar reasons, it seems at least possible (if not likely) that OO may
be rejected for inclusion in a future GNOME release (note that Mozilla
even uses GTK+ underneath its custom chrome, and is smoothly embeddable
in GNOME apps through the use of a Bonobo control - this was not deemed
sufficient to be considered a part of GNOME).

3) Sun has expressed a very strong desire for OpenOffice to be a part of
GNOME in particular as a part of their new GNOME thrust, to the extent
that they considered the Foundation's (to be anally precise, at the time
it was the steering comittee) adoption of OpenOffice as part of GNOME
Office critical to Sun's own adoption of GNOME. Sun says, "As part of
its formation, the GNOME Foundation has announced that it is adopting
OpenOffice.org -- the StarOffice productivity suite that Sun is making
available to the open source community -- as the core of the office
productivity software for GNOME Office." The point is...Sun seems to
honestly care that OO be a part of GNOME...lets encourage this!

4) Sun recognizes the value of uniformity in platform and interface a
couple sentences later, and lauds GNOME as the herald of such. "...GNOME
delivers a more uniform and consistent user experience across
traditional PC and workstation systems as well as next-generation
Internet appliance devices. This uniformity dramatically improves ease
of use." ie. Sun seems to recognize inherent worth in providing users
with a consistent interface, that interface being the GNOME interface.

5) On a technical note, after some perusal of the OO VCL codebase (their
existing system abstraction library for widgets and graphics devices)
and reading all the previous fleeting discussions related to this in the
openoffice.org mailing list archives I could find...it seems that
conversion to GTK+, while by no means trivial, also would not require
large rewriting of the codebase aside from VCL itself. I could be dead
wrong about all of this, it is a fly-by-night assessment, but I don't
say this entirely without grounds :-) Presumably the existing VCL code
would be kept for use on Win32, non-GTK-X, OS/2, Macintosh, etc. There
may be a speed penalty since this would mean adding yet-another-layer of
complexity between a user event and the response reaching the display
hardware, however I wonder if this negative effect would be somewhat
nullified, as I suspect GTK+/GDK is better optimized and uses hardware
acceleration better than their in-house-non-system-specific system. 

To be frank, the existing OO method isn't ideal for creating seemless
Windows applications either. If you run OO on a WindowsME machine you
can already tell that they are not using native Win32 widgets, it still
looks like the application would under Win95. I know it could be painful
technically, but it might be in Sun's interests to consider actually
rewriting VCL to allow for system specific widget backends. This was
used and did have problems in SO 3.x (apparently motif had some really
stickly focus interactions with their own focus system), but I think the
technique has been successfully used with other applications in the
past. It would *also* remove the burden of having to maintain a widget
system from the OO developers in the long run, freeing those developers
to contribute to the core functionality of OO itself. Is it really
efficient for an application to maintain its own widget system?!?

Thus,

6) Sun would probably be interested in funnelling some development
effort into making OO use GTK+ and gnome-libs if it increased its
adoption into GNOME. They really want to be a part of GNOME, many of us
want them to be a part of GNOME, but it seems some work needs to be done
before everyone would happily consider them as such. It could well be
that this integration is already planned but such information hasn't
spread well outside of Sun, that we're bringing new information to the
table they hadn't considered...or any other number of trivial
communications problems, esp. since the OO team probably needs to be
focused in the short term on reaching par with the old StarOffice
codebase (things had to be removed to release the cobe as free software)
and doing a release at that point.

Dark omens from the pit of my cynical soul:

On a more cynical note, it is worth noting that although
www.openoffice.org originally made mention of GTK+ and Bonobo in its
FAQs (in a few places) concerning future directions...according to their
CVS tree these notations were explicitly sanitized, which causes me some
concern (idle speculation: perhaps this means Sun was planning to make
OO a serious part of GNOME but later arrived at some internal decision
not to, and updated the web site to avoid furthering this notion to the
extent that it wasn't already disseminated). I have also at times
received the mild impression that parts of the StarOffice team acquired
with Star Division have not been throughly sold on the idea of GNOME, or
more generally, on not having OO be a desktop environment unto itself.

ACTION REQUEST:

Would Havoc or Maciej be willing to bring some of these thoughts (or
other related community ruminations) to the attention of John Heard at a
GNOME foundation board meeting? (I'm rather attached to this message
itself and wouldn't mind its explicit use...though I doubt the arguments
are nearly as convincing as they seem to its author at 5:30am ;-)

-the sleepy seth