Evolution release, major regressions, and GNOME release process

[Andrew Montalenti <ajm pixelmonkey org>]

I recently upgraded from Ubuntu Hardy to Ubuntu Intrepid.  Most of my
GNOME software worked well upon upgrade, there were even some
improvements.  However, a major piece of GNOME software that I use every
single day and that is important to both my personal life and business
stopped functioning correctly: Evolution.

>From 2.22 -> 2.24.3, Evolution redesigned its backend data store and
unleashed this onto users as a "stable release" with disastrous effects.
If you look at the evolution-list mailing list, you'll find lots of
discontent.  Basic things that existed in Evolution 2.22 -- basic
features -- no longer work.  For example:

* Search folders or searches including "unread message" status does not
work (regression)

* Declaring a search folder of a search folder (vfolder of a vfolder) no
longer works (regression).  For me and many others on the list, this
essentially meant that all of our vfolders stopped working altogether,
meaning that we had to reorganize our mail in a different way.
(regression)

* Unread message counts are incorrect vis-a-vis the reality.  For
example, every time I send a message, my inbox's unread message count
goes up by 1.  This despite the fact that there are no unread messages
there.  Others report even larger divergences. (regression)

* The "Unmatched" VFolder no longer exists.  (regression)

It turns out that these fixes still have no been committed even in Evo
2.26, released in March.  Brian J. Murrell on the evolution-list has
been indignant about this.

Why am I contacting the GNOME release and marketing teams?  Because I
consider Evolution to be one of the core pieces of software that GNOME
offers, and one that the release team should carefully watch when
declaring new "stable" releases of GNOME.

I think the fact that Evo was allowed to be released at this new version
with so many regressions is really a sad state of affairs, and suggests
that you need to reconsider your release process.  Evo worked fine at
2.22, and I see very few improvements between 2.22 and 2.24, only
regressions.  That suggests that someone on the Evo team thought it was
a good idea to "redesign the internals" without really committing to
what that entailed -- releasing the redesigned version to a small "beta"
community before declaring it to be the "stable version" released to
thousands of production users and including in all major distributions
as the latest and greatest STABLE software from GNOME.

I don't know what can be done about it now, but this reflects very badly
on GNOME for me.  I'm a long-time GNOME user (~ 10 years), and in recent
years as my computer has become more and more integral to my livelihood
(as a small business owner and software engineer), I have become more
and more hesitant about upgrading to the latest GNOME versions.  That's
why I was still running Ubuntu Hardy as of a week ago.  At least it
worked.  I upgraded to Intrepid only because Jaunty is right around the
corner, and I figured this would at least represent a "stable" snapshot
of software.  I guess I was wrong.

What does it say when some of your most committed users (users who have
hacking credentials -- I know C, GTK+, and GObject!) are hesitant to
upgrade to your latest stable releases?  Even worse, what does it say
when their hesitancy is justified?

I really want the best for GNOME and want to see this situation improve,
but I don't think it will unless someone with serious credentials within
GNOME takes a strong stand.

Andrew