Re: Empathy account UI freeze break request

[Sjoerd Simons <sjoerd simons collabora co uk>]

On Mon, Aug 24, 2009 at 02:42:05PM +0200, Xavier Claessens wrote:
> 1) UI changes are not really bound to the mc5 migration. They were made
> in the same branch for facility, but I think we can do UI changes
> without moving to mc5 and vice-versa. Probably too late to "unbound"
> those 2 changes now...

In theory they are seperate, in practise they are completely tied together.
Also they would never have been unbound as there is no point on developing
Empathy on mc4 further.

> 2) There were not that much testing. This morning I had 2 kinds of
> critical crashers (they are probably fixed now): When first starting
> empathy, and we opening any conversation.

This was because you triggered a bug that already existed in the current master
version of empathy. You just happened to not trigger it there, also note that
this is already fixed.

Me and several others have been running it for quite a while without any issues
and we have several people testing it quite heavily now and for the upcoming
weeks as we stabelize things for Gnome 2.28

> 3) I totally agree with that assistant, it is great for new users.
> Though, I don't like at all changes made to the account dialog (purely
> personal opinion).

We made this in consultation with some UI experts, it's not fully there yet but
it's definately in the right direction.

> 4) What's the status of "Share my desktop" integration? I think it
> doesn't work with mc5, do we have a patch now? The irony is that's the
> only integration we have for GNOME 2.28 and it works fine with mc4...
> Even if that's because of ugly hacks, that becomes much cleaner with
> mc5.

That works with some small changes. There was a small part of share my desktop
that poked MC4 to get some information, which triggered an issue. In the case
of MC5 it gets the necessary information right away and doesn't even has to
poke mission-control which should actually simplify the code quite heavily.

Someone is working on doing this right now.

The other tubes applications we've tested worked fine without any changes (for
example gnome-sodoku which was done during the current SOC)

> 5) Lots of protocols are now duplicated in the create account UI.
> Because we can connect them (jabber, irc, salut, msn) using
> telepathy-haze (using libpurple) or the dedicated telepathy connection
> manager. The user has no clue how to make a decision. And what "haze"
> could mean?!?

That's being changed as we speak, it was there for development purposes only.

> 6) "Jabber" and "GTalk" got replaced by "XMPP" which is totally unknown
> to normal users (pidgin does the same error). We already rejected the
> change from Jabber to XMPP in a bug report some months ago, and I think
> we should keep a gtalk entry even if it's the same as Jabber...

That will be fixed before the G2.28 release.

> 7) Is gnome-keyring integrated in MC5 now? I see them in plain text in
> ~/.mission-control/accounts.cfg.

This is totally up to mission-control, we've got a branch of mission-control
ready which stores them in the keyring instead.

> IMO, that's good material for branching Empathy for GNOME 2.29 right
> now, to not take the risk of that enormous code change,

That enormous code changes has been hatched out for a while now while you
weren't paying attention. Also we have several people testing and improving it,
while the MC4 master branch of empathy is complete a dead end.

> and UI changes that only a few hackers tested, with no community feedback.

We didn't have community feedback, but we had some UI experts working with us
which is in my opinion much more useful.

> Empathy 2.28 is going to be the first version integrated by default by
> lots of popular distro, we can't take any risk of regressions.

That's one side of of reasoning, the other thing is that we want to push
things forward now, so that all these new users will not be stuck on MC4 and
then needing to move them over to MC5 in the next gnome release when are
user-base is a lot bigger. Also we'd like to support the G2.28 Empathy release
nicely, but if this branch isn't merged there is no real point for us in doing
that.

Also we don't want to stay in the twilight area in which we're now where we
have need to warn tubes applications writes all the time that they'll have to
change when MC5 comes. The current Empathy mc5 branch will support the old way
of empathy launching tubes apps, which we're planning to scrap completely for
G2.29.

This means that the authors of the current applications have a nice transition
period and it also means that new tubes app writers can target MC5 directly and
things will work on a stable gnome as well.

> If it was proposed 1 or 2 months ago, I would have given a clear +1 for
> merging. But, now, I let the decision to the release team...

This work has been going on for 1 or 2 months now, the fact that you didn't
take a look even though you promised to help out doesn't mean it's not ready :)

As i said earlier we've got some people smoketesting this branch quite heavily
to shake out some more bugs and the only real thing that we've seen that breaks
is the share-my-desktop in vino, which we already have some working on to fix
it (and he should actually be able to simplify the current code a lot).

Basically if this branch is not merged, there is no benefit for us to keep
working on the G2.28 of Empathy. Which would be a big loss for G2.28 imho.

  Sjoerd
-- 
There are no winners in life, only survivors.