Re: Issuing a press release about GNOME 3

[Brian Cameron <Brian Cameron Sun COM>]

Luis:

> I hate to dump on well-meaning people like those of you on r-t, but
> this 3.0 plan is, hands-down, a terrible idea, on a lot of levels, and
> Dave's points here- why are we focusing on API? what are we signaling
> to users?- begin to highlight why. The fact that the very first
> sentence of http://live.gnome.org/GNOME3 is wrong: "GNOME 3 is needed
> as GTK+ 3 will happen." is just not a good sign. (There are ways to
> educate developers about API/ABI change besides major version numbers
> of the desktop, so GTK3 need not force GNOME3.)

I disagree.  Since GTK3 is going to force us to make the GNOME stack
parallel installable, this is a reasonable time for us to do a break
with the past.  This will likely involve removing deprecated libraries
from the stack.  We could still support them, but at least ensure that
the "official" GNOME stack doesn't use them.  This way distros can drop
them if they want.  Providing a cleaner, more powerful, and more stable
development platform is a good thing.

> In short, I think you're letting minor technical considerations (and
> perhaps perceived pressure from KDE?) set out an agenda, rather than
> making the user and improvements for the user set the agenda, and I
> think that is exactly backwards, screwing us up with users,
> developers, and the media.

I do agree with you.  I think it would be a shame if GNOME 3.0 only
provided developer Platform enhancements with no focus on the user
experience or usability.

I would hope that a part of GNOME 3.0 planning would include some
discussion about what usability improvements should be a part of the
release.  I'd think we should be investing in doing things like
usability studies and starting forum discussions to get a better handle
on what we could/should be doing in this area.

> In more detail:
>
> First, from a user perspective: how am I supposed to understand what
> kind of change has gone on here? The change in major number is
> supposed to indicate radical change. That is what version numbers do.

That is one aspect of what version numbers do.  Version numbers also
highlight information about interface stability.  That said, it would
be best, obviously, if GNOME 3.0 also included some real usability
improvements as well.  Since GTK3 won't be available for another year,
and GNOME3 would likely lag another 6-12 months, I'd think we should
have time to identify and implement some significant usability
improvements.  No?

For example, why does the GNOME desktop have separate panel applets and
things like gDesklets.  Why can't I just drag the clock applet onto my
desktop and get a gDesklet-like thing.  I know this is the sort of
improvement that Calum Benson has often talked about.

Also, we could work to finish-up and polish composite and Clutter
integration with the platform making 3.0 have the sort of new user-flash
that would generate some excitement, I'd think.

In short, I think there are things in the pipeline that could generate
the sort of user-focused features you suggest.  However, I'd think we
would need to invest some energy working with our users to make sure
it integrates with the appropriate polish to justify a 3.0 release.

> Second, from a developer perspective: I understand the need to
> indicate to developers that an API/ABI change has occurred, but if we
> need to, that is why we have a platform/desktop split- change the
> version number in the platform. Changing the desktop version without a
> clear vision/agenda, *especially* combined with new API/ABI + porting,
> is an invitation to architecture astronautics and unnecessary churn.

It could make a lot of sense to make the GNOME Platform 3.0 and still
leave the desktop at 2.0.  However, this would mean that all desktop
programs would need to continue to be backwards compatible with 2.0
moving forward.  So desktop programs couldn't make use of any new
GTK3-only interfaces until the desktop moves to 3.0.  From the GTK3
plan, it doesn't sound like GTK3 will initially have many new features.
So it might not be a problem if the desktop version were to get
bumped a release or two after the platform.  This would offer some
additional time to make whatever changes to the desktop we feel are
necessary to justify the 3.0 version bump for the desktop.

In short, I think you are right that we need some further planning to
make sure that a GNOME3 release is successful.

> Finally, from a media perspective: the reason GNOME 2.0 was a success
> in the Linux media, and the reason KDE 4.0 has been a failure, is that
> GNOME 2.0 had a clear, persuasive story around it: simplification and
> usability.

I think another major factor is that the GNOME Platform is LGPL whereas
KDE is GPL.  Therefore GNOME is much more friendly to third-party ISV's.

Brian