Stricter policy when including BIG modules (was Re: Evolution 2.0 and GNOME 2.6)

[Danilo Segan <danilo gnome org>]

Hi Murray,

Murray Cumming <murrayc murrayc com> writes:

> GNOME's quality control means that there is always a risk that a module
> might be punted or reverted to a previous version.

Of course.  But I mentioned that special care should be taken with
such modules like Evolution: it's a huge program itself, it has over
3900 messages to translate.

What I'm arguing for is not to let it slip into the regular release
cycle if it's not on par with the rest of Gnome.

>
> The translations are not lost. People will still get a lot of benefit
> from them, and it will help greatly with GNOME 2.8. It's a shame if
> people's only GNOME 2.6 translators credits would be for Evolution, but
> I hope that people are translating evolution because they want to use it
> in their language rather than to see their name in a list. 
>

You missed my point.  Gnome 2.6 is coming in early March. That means
around one month for translators to complete it.  That means that
Evolution translation could have waited until after Gnome 2.6 was
released.  And it also means that many translators could have
actually gone closer to getting Gnome 2.6 translated, if they didn't
spend their time on Evolution NOW.  So, it's a waste of work in a way
(not to mention that this means that Evolution won't be following
string freezes, and that there will be *real* waste, because messages
will change before actual release as well).

Evolution has been postponed for 3 months, it was not merely removed
from the release.  In those 3 months, only ones who will benefit from
these translations are testers and early accepters -- a minority.

(btw, I mentioned language names/teams being "in a list", not people
names, at least that's what happened in 2.4 RN -- this is entirely
different motivation from what you're implying)

> I understand, and I think motivation is very important, but it's not the
> most imporant thing and this is not the most imporant way to motivate
> people.

Well, I cannot debate on which motivations are "important" and which
are not (I consider them all equally important, but that's just my
opinion), but I'm glad that you agree that this is an issue.

Gnome is also using a list of supported languages for marketing
(read the thread a while back about KDE having better support for
more languages -- now here's the culprit, I say).  

That's why I repeat: this has a negative impact on Gnome community as
a whole.

>> So, what I ask for here:  establish a policy that any significant
>> module (in terms of work for other Gnome subprojects, such as
>> documentation, UI review, translation, etc.) must guarantee not to
>> withdraw once final module list is announced.
>
> No way. We must have quality control. For instance, Totem+gstreamer
> wasn't ready so it wasn't in GNOME 2.4, and that' good.

Yeah, and go back to my point you quote: "significant module (in
terms of work...". You can even add Totem now to Gnome 2.6 and nobody
would complain that loud: it's a consisted of 333 messages. And now go
back and read Evolution count: 3900+ (with E-D-S, gtkhtml, gal, close
to 5k).  I want to say that Evolution being in and out is not same as
doing it with Totem.

In simple words: for such influential modules such as Evolution, be
more strict and careful when adding it to the core (that was the
intent of my proposal, not to bind ourselves on unready modules --
when there's such a penalty, one tends to be more careful about
choices, I would say).

I'm not pointing fingers at anyone, for I consider this a problem for
all of us.  But I want us to resolve that and admit a mistake.
There were hints all over the place of Evolution not being ready (my
memory may be failing me, but I think both Jeff and you doubted it
will be ready in time),  but it seems to me that release team, and all
other participants in discussions actually went by what they *want* to
happen, and not what's realistic (I don't exclude myself here, I
wanted it in probably as much as everybody, and that's why I never
complained about it).

I18N issues were actually brought up early in the process -- this was
done by Christian Rose when Evolution inclusion was discussed. So,
with all those precautionary notes, this still happened.

And that's why we need policy, so we'd be able to decide without
doubt if module is to be included or not. It can be as simple as: if 
we're not sure it's going to be ready and meet Gnome quality
standards, and it greatly affects work of Gnome subteams, then don't
include it.

As we always remind others: another release comes in 6 months, so
nothing to complain about if it's not included right away.

I'm pointing out at the problem in the release process.  It's not a
terrible problem, but I think it needs to be fixed, not neglected.
I'm offering a solution as well, it's just up to the community to
recognize the problem, and choose a solution (not necessarily mine, of
course).

Cheers,
Danilo

PS. As Blaise Pascal would put it: excuse me for such a long letter,
but I lack the time to make it shorter.