Nokia released Mission Control

As announced on telepathy ML [1], nokia released their MC. This means
gossip will need a big rework if we want to take advantage of it.
Basically the MC is a layer on top of telepathy that manage accounts.
This allow to make a very small separate UI program for each aspect of
an full IM client. For example we can have one program to manage contact
list, one for private chat, etc.

I'm posting here to get opinions of what will gossip become. I think it
will need lots of work and re-factorisation in order to get it good.

In a first step I think we should keep gossip in one piece, or at least
one tarball. I didn't read the whole libmissioncontrol API but I guess
it will mainly need rework in GossipAccount, GossipAccountManager and
related UI.

The big question here is: Do we keep GossipJabber backend ? I think it
won't be possible to keep a no-telepathy backend if we want a good MC
(and desktop) integration, or at least it will need more work and I'm
not sure there is still developers that have time to take care of this
(old) backend.

Xavier Claessens.

[1]
http://lists.freedesktop.org/archives/telepathy/2007-March/000471.html

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Tomasz Sterna]

Dnia 10-03-2007, sob o godzinie 12:42 +0100, Xavier Claessens
napisał(a):

> Basically the MC is a layer on top of telepathy that manage accounts.
> This allow to make a very small separate UI program for each aspect of
> an full IM client. For example we can have one program to manage
> contact
> list, one for private chat, etc.
>
> The big question here is: Do we keep GossipJabber backend ? I think it
> won't be possible to keep a no-telepathy backend if we want a good MC
> (and desktop) integration, or at least it will need more work and I'm
> not sure there is still developers that have time to take care of this
> (old) backend.

Could we use MC to manage not only telepathy but also Gossip (and other)
accounts? Perhaps extending MC to do it and submitting it back.

I'm no big fan of telepathy. It's nice, yes.
But it also bounds us. We cannot do and experiment with things that
telepathy does not handle (yet). It also depends us hardly on something
we don't fully control.

Gossip is a nice Jabber (XMPP in future maybe) client. One of the
nicest. Dropping backend support in favor of telepathy would make it
just another telepathy frontend.


--
Tomasz Sterna
Xiaoka Grp.  http://www.xiaoka.com/

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

On dim, 2007-03-11 at 11:34 +0100, Tomasz Sterna wrote:

> Dnia 10-03-2007, sob o godzinie 12:42 +0100, Xavier Claessens
> napisał(a):
> > Basically the MC is a layer on top of telepathy that manage accounts.
> > This allow to make a very small separate UI program for each aspect of
> > an full IM client. For example we can have one program to manage
> > contact
> > list, one for private chat, etc.
> >
> > The big question here is: Do we keep GossipJabber backend ? I think it
> > won't be possible to keep a no-telepathy backend if we want a good MC
> > (and desktop) integration, or at least it will need more work and I'm
> > not sure there is still developers that have time to take care of this
> > (old) backend.
>
> Could we use MC to manage not only telepathy but also Gossip (and other)
> accounts? Perhaps extending MC to do it and submitting it back.

No, the MC is a layer on top of telepathy, it depends completely on it.
The big difference is an account connected using telepathy can be
accessed via dbus, so we can have many clients that use the same
connected account, that's fundamental to integrate IM in the desktop
which is the goal of MC.

> I'm no big fan of telepathy. It's nice, yes.
> But it also bounds us. We cannot do and experiment with things that
> telepathy does not handle (yet). It also depends us hardly on something
> we don't fully control.

We depend on Gtk, should be recode it to be sure to fully control it? Do
one think but do it well, gossip is an IM client not an IM protocol
backend. I totally convinced that telepathy is THE solution for all IM
UI.

> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
> nicest. Dropping backend support in favor of telepathy would make it
> just another telepathy frontend.

No, gossip is yet-another-jabber-client with nothing really special,
moving to telepathy opens the door to a new world of desktop integration
and would make gossip really interesting.

For me it's simple, move to telepathy or die. Traditional clients like
gaim which re-implement all protocols has no future, it can't be easily
integrated with other desktop applications and has too many code
duplication with other traditional IM clients.

Take a simple example in gossip the telepathy backend supporting VoIP
and all possible IM protocols with almost the same features as the
jabber backend takes 6923 lines of code, gossip's jabber backend that
can't do msn/irc/etc and without VoIP takes 13691 lines of code... Using
telepathy we support far more with the half of code, that's the proof
there is a lot of code duplication in traditional IM clients.

Xavier.

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Mikael Hallendal]

Xavier Claessens skrev:

Hi,

First of all, I found your mail a kick in the face to those that have
spent years on trying to create a nice UI.

A nice UI is in my opinion what was lacking (and still to large extent
is) in other Jabber clients, so in that sense Gossip is not
"yet-another-jabber client with nothing really special".

>> I'm no big fan of telepathy. It's nice, yes.
>> But it also bounds us. We cannot do and experiment with things that
>> telepathy does not handle (yet). It also depends us hardly on something
>> we don't fully control.
>
> We depend on Gtk, should be recode it to be sure to fully control it?

This is just silly and I hope you were trying to be funny.

>> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
>> nicest. Dropping backend support in favor of telepathy would make it
>> just another telepathy frontend.
>
> No, gossip is yet-another-jabber-client with nothing really special,
> moving to telepathy opens the door to a new world of desktop integration
> and would make gossip really interesting.

Gossip is already really interesting to a lot of people. But yes, I do
think that having support for other IM protocols will make it more
interesting for more people. However, keep in mind that users don't care
much whether it's done the gaim way or the Telepathy way.

> For me it's simple, move to telepathy or die. Traditional clients like
> gaim which re-implement all protocols has no future, it can't be easily
> integrated with other desktop applications and has too many code
> duplication with other traditional IM clients.

I'm quite sure Gaim won't roll over and die (and I sure hope they
won't). Gaim has a huge user base and as long as Gossip/Telepathy or
whatever other solution you choose doesn't provide something better
there is little that will make people switch.

As for the tasks at hand. From what I understand the idea about
Telepathy is that "IM" will be split between several clients all sharing
the connection through Telepathy. That is, one application for a contact
list, another for a chat dialog (that could be started from any
application on the desktop), one for voip and so on.

Keeping that in mind I'm not sure the best way to go forward is to
completely rework Gossip to become those different parts. Or that we
should first try to jam it all into Gossip just to later split it up.

Maybe you should attack the problem from the other way. Take parts from
Gossip and create separate small apps?

In the future, keep in mind that there are those in the Gossip community
that are only interested in the Jabber protocol and couldn't care less
about MSN, ICQ etc. For them it's not Telepathy or die, in fact, it's
likely that Telepathy is only a drawback as it makes the UI more complex.

My interest in Telepathy has nothing to do with multi protocol support
but the possibility to make XMPP more integrated in the desktop. Whether
this is done through Telepathy or a desktop XMPP-server matters little
to me (from a user point of view).

Regards,
  Mikael Hallendal

--
Imendio AB, http://www.imendio.com/
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Martyn Russell-3]

Tomasz Sterna wrote:

> Dnia 10-03-2007, sob o godzinie 12:42 +0100, Xavier Claessens
> napisał(a):
>> Basically the MC is a layer on top of telepathy that manage accounts.
>> This allow to make a very small separate UI program for each aspect of
>> an full IM client. For example we can have one program to manage
>> contact
>> list, one for private chat, etc.
>>
>> The big question here is: Do we keep GossipJabber backend ? I think it
>> won't be possible to keep a no-telepathy backend if we want a good MC
>> (and desktop) integration, or at least it will need more work and I'm
>> not sure there is still developers that have time to take care of this
>> (old) backend.
>
> Could we use MC to manage not only telepathy but also Gossip (and other)
> accounts? Perhaps extending MC to do it and submitting it back.
>
> I'm no big fan of telepathy. It's nice, yes.
> But it also bounds us. We cannot do and experiment with things that
> telepathy does not handle (yet). It also depends us hardly on something
> we don't fully control.
>
> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
> nicest. Dropping backend support in favor of telepathy would make it
> just another telepathy frontend.

Thank you, and that is one of the the aims of the project. To be nice
looking, easy to use, etc. If you feel that way, we should be doing
something right - and we hope to continue this.

I have been sitting on the fence for a long time with Telepathy. I have
no doubt in its merits and abilities but I am really just trying to make
up my mind, should we drop the Jabber backend for Telepathy. What makes
this decision hard is, without Telepathy Gossip is useless in that case.
Say we want to put Gossip on Windows - can we do that with Telepathy? Or
even on MacOS?

What we or I have been doing wrong all these years is basically
abstracting the whole contacts/presence/etc libgossip layer to work with
multiple protocols, Mikael originally started doing the abstraction to
make the code simpler to maintain. I think it wasn't a good idea and it
hasn't done much for us at all, most of that code will be redundant if
we choose Telepathy and drop the Jabber backend, and likewise, if we
decided to do the same but for Jabber. This is my fault, I think I
thought I could do it nicely and have backends written to just plugin.

When it comes to Telepathy, I am afraid that I agree whole heartedly
with Mikael, really I think that it would be best in many ways to just
write new smaller apps from scratch and perhaps to begin with have them
all in an umbrella tarball called Gossip. In my UI experience, the most
important thing is keeping the complicated parts out of the way of the
UI, i.e. having the telepathy operations somewhere and keeping the UI
code clean. We can still do that I think with the work that Xavier has
been doing.

So I have been thinking about how to reuse all the experience we have
gained and things we have fixed/implemented in Gossip to make sure we
don't make those mistakes all over again. What I have considered is
putting common widgets into a library which we can call on and use where
necessary. Short of that, it should be very simple to just create simple
stand alone applications to do what we want.

What I am still pondering over is, do we keep the Jabber backend and
make Gossip able to build in the event Telepathy doesn't exist/isn't
installed/etc. I don't think we have the man power realistically to do
it and will probably end up dropping it, which will be a shame.

>From my experience with the patches and comments on #gossip (IRC) it
sounds like Telepathy has been carefully thought out and has some good
advantages to libgossip in the way things are managed and the events
that are emitted. We always seem to be playing catchup in libgossip to
fix things up to work nicely with Telepathy, and that is where we spend
a lot of time at the moment, which shouldn't be the case. We should be
writing nice, cool UIs with highly usable interfaces and nice desktop
integration - but have we had time for that? NO.

As for the MC release, that's is all good, but this really just
highlights ANOTHER problem I have seen coming for some time and not
really known how to deal with. That is that we have 'n' places where we
store our details on the desktop and there is no ONE place. Well at
least we have the gnome-about-me dialog which tries to solve that any
way and we are currently doing work in Gossip with Fillipo to try to
make the intergration better. However, the accounts information in
Gossip is something that you can configure in gnome-about-me, Gossip,
(does evolution have this input somewhere I can't remember?), and now
MC. I am not sure that using MC from the Nokia device is the best way
forward, but then I haven't looked at it enough yet to comment properly.

This week I will be discussing all these things with Mikael and Richard
and come up with a clear direction for Gossip and roadmap.

/martyn-dump :)

--
Regards,
Martyn
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Jerome Haltom]

> As for the MC release, that's is all good, but this really just
> highlights ANOTHER problem I have seen coming for some time and not
> really known how to deal with. That is that we have 'n' places where we
> store our details on the desktop and there is no ONE place. Well at
> least we have the gnome-about-me dialog which tries to solve that any
> way and we are currently doing work in Gossip with Fillipo to try to
> make the intergration better. However, the accounts information in
> Gossip is something that you can configure in gnome-about-me, Gossip,
> (does evolution have this input somewhere I can't remember?), and now
> MC. I am not sure that using MC from the Nokia device is the best way
> forward, but then I haven't looked at it enough yet to comment properly.

I a potential solution to this is to integrate a single MC into
Telepathy itself. Or at least the configuration parts of it. Convince
the telepathy people that that is the way forward. Basically said MC
just keeps whatever accounts you tell it to online, and provides a
single account store in .telepathy/ or some such (gconf).

Gossip then, perhaps unfortunately, perhaps fortunately, would have the
UI to do that manipulation and instruct the MC to log on and off. The UI
that Gossip already has would be nice for this.

You can then break Gossip apart into n-teenth shared widgets for
interacting with Telepathy back-ends.

All of which is a lot of work.

Good idea, bad idea? Got me. I do suspect that if Gossip doesn't do it,
something else eventually will. I think that's basically just the
direction our desktop (and other desktops) are heading in. I also think
Gossip has a bunch of this work already done, and it would benefit us
all of it was used instead of reinventing it again.

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

On dim, 2007-03-11 at 16:17 +0100, Mikael Hallendal wrote:
> Xavier Claessens skrev:
>
> Hi,
>
> First of all, I found your mail a kick in the face to those that have
> spent years on trying to create a nice UI.

Sorry if I offended you. Be sure I really like gossip, for me it has the
best UI ever, it's very well coded and has a good design, really a very
good software! I didn't say it's actually bad, what I want is to push it
to a even higher level, to something totally new that no other IM client
has.

For example gossip is (will be) the first IM client to support VoIP with
jabber/gtalk thanks to telepathy. With that kind of think we are no more
on the state of a simple jabber client, we do more!

> A nice UI is in my opinion what was lacking (and still to large extent
> is) in other Jabber clients, so in that sense Gossip is not
> "yet-another-jabber client with nothing really special".

Totally agree, the UI is very nice! But I'm not talking about UI, I talk
about the backend. Otherwise I wouldn't work to implement telepathy into
gossip but I would start a new program designed from the ground to work
with telepathy. I want to keep the ui and work only on the backend.

> >> I'm no big fan of telepathy. It's nice, yes.
> >> But it also bounds us. We cannot do and experiment with things that
> >> telepathy does not handle (yet). It also depends us hardly on something
> >> we don't fully control.
> >
> > We depend on Gtk, should be recode it to be sure to fully control it?
>
> This is just silly and I hope you were trying to be funny.

Dependencies are made to be used and to avoid duplicate work. That's
true for gtk (to build UI) and for telepathy (to build IM)!

> >> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
> >> nicest. Dropping backend support in favor of telepathy would make it
> >> just another telepathy frontend.
> >
> > No, gossip is yet-another-jabber-client with nothing really special,
> > moving to telepathy opens the door to a new world of desktop integration
> > and would make gossip really interesting.
>
> Gossip is already really interesting to a lot of people. But yes, I do
> think that having support for other IM protocols will make it more
> interesting for more people. However, keep in mind that users don't care
> much whether it's done the gaim way or the Telepathy way.

I know, but the telepathy way opens the door of many more than just a IM
client, it provides a solution for desktop integration of all IM
concepts. I think users care about that.

> > For me it's simple, move to telepathy or die. Traditional clients like
> > gaim which re-implement all protocols has no future, it can't be easily
> > integrated with other desktop applications and has too many code
> > duplication with other traditional IM clients.
>
> I'm quite sure Gaim won't roll over and die (and I sure hope they
> won't). Gaim has a huge user base and as long as Gossip/Telepathy or
> whatever other solution you choose doesn't provide something better
> there is little that will make people switch.

True, gossip don't provide something better, but a telepathy/MC aware IM
client with a good desktop integration does provides something totally
new and far better.

> As for the tasks at hand. From what I understand the idea about
> Telepathy is that "IM" will be split between several clients all sharing
> the connection through Telepathy. That is, one application for a contact
> list, another for a chat dialog (that could be started from any
> application on the desktop), one for voip and so on.
>
> Keeping that in mind I'm not sure the best way to go forward is to
> completely rework Gossip to become those different parts. Or that we
> should first try to jam it all into Gossip just to later split it up.
>
> Maybe you should attack the problem from the other way. Take parts from
> Gossip and create separate small apps?

I'm not sure for the best way here. First of all I really don't want to
fork gossip UI and most of libgossip, if we split gossip into pieces I
want to be sure people won't keep old gossip and continue work on it.
That's why I prefer to keep gossip into one piece atm, or at least one
tarball/project.

> In the future, keep in mind that there are those in the Gossip community
> that are only interested in the Jabber protocol and couldn't care less
> about MSN, ICQ etc. For them it's not Telepathy or die, in fact, it's
> likely that Telepathy is only a drawback as it makes the UI more complex.

I know, actually we support MSN/Salut/Jabber/IRC (and more?) using
gossip-tp and the UI keeps clean and almost unchanged. I think
multi-protocol does not force the UI to be more complex.

> My interest in Telepathy has nothing to do with multi protocol support
> but the possibility to make XMPP more integrated in the desktop. Whether
> this is done through Telepathy or a desktop XMPP-server matters little
> to me (from a user point of view).

Telepathy provides both advantages: desktop integration and multi
protocol support. For me desktop integration is the best part and can
begin if we adapt gossip to work with the MC.

> Regards,
>   Mikael Hallendal

Thanks for your comments,
   Xavier Claessens.

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Martyn Russell-3]

Xavier Claessens wrote:
> On dim, 2007-03-11 at 11:34 +0100, Tomasz Sterna wrote:
>> I'm no big fan of telepathy. It's nice, yes.
>> But it also bounds us. We cannot do and experiment with things that
>> telepathy does not handle (yet). It also depends us hardly on something
>> we don't fully control.
>
> We depend on Gtk, should be recode it to be sure to fully control it?

That's a really bad comparison. What Tomasz is saying, is basically he
can take Gossip and install it stand alone on his machine - he KNOWS he
will need dependencies like GTK+ etc, but that comes for free in 99.99%
of the cases, having Telepathy doesn't and that is quite important to
remember here. If Telepathy was part of Gnome and could easily be used
on other Operating Systems, then I think it would make the transition
and idea more comfortable for everyone. At the end of the day users want
to just install one thing, not a list of things. This is something the
developer/maintainer has to consider in such situations.

> Do
> one think but do it well, gossip is an IM client not an IM protocol
> backend. I totally convinced that telepathy is THE solution for all IM
> UI.

Actually, that is wrong. Currently Gossip is an IM client with an IM
backend and support for integration with an IM subsystem.

>> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
>> nicest. Dropping backend support in favor of telepathy would make it
>> just another telepathy frontend.
>
> No, gossip is yet-another-jabber-client with nothing really special,
> moving to telepathy opens the door to a new world of desktop integration
> and would make gossip really interesting.

I'm sorry Xavier, but I have to agree with Mikael here. That is a bit of
a kick in the bollocks. If you feel that way about Gossip, why work on
it? Why work on it in the first place if it is the same as all the rest?
I am sure you don't agree with that comment especially with all the hard
work you have put in.

> For me it's simple, move to telepathy or die. Traditional clients like
> gaim which re-implement all protocols has no future, it can't be easily
> integrated with other desktop applications and has too many code
> duplication with other traditional IM clients.

Really it doesn't matter what GAIM does. I consider it far to broken for
me to even worry about. But that is my personal view, I wouldn't be so
hasty to condemn GAIM, it is still used by a LOT of people as Mikael says.

> Take a simple example in gossip the telepathy backend supporting VoIP
> and all possible IM protocols with almost the same features as the
> jabber backend takes 6923 lines of code, gossip's jabber backend that
> can't do msn/irc/etc and without VoIP takes 13691 lines of code... Using
> telepathy we support far more with the half of code, that's the proof
> there is a lot of code duplication in traditional IM clients.

You are comparing apples to pairs here. I mean really, think about it,
the Jabber backend has to interface to Loudmouth (which Telepathy does).
So you are one layer higher with the Telepathy backend in Gossip. All
the hard work by Daf, Robert, et al is done for you, all you have to do
is connect a few signals, send a bit of data, remember some information,
bla bla, it is a lot easier trust me.

--
Regards,
Martyn
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Martyn Russell-3]

Jerry Haltom wrote:

>> As for the MC release, that's is all good, but this really just
>> highlights ANOTHER problem I have seen coming for some time and not
>> really known how to deal with. That is that we have 'n' places where we
>> store our details on the desktop and there is no ONE place. Well at
>> least we have the gnome-about-me dialog which tries to solve that any
>> way and we are currently doing work in Gossip with Fillipo to try to
>> make the intergration better. However, the accounts information in
>> Gossip is something that you can configure in gnome-about-me, Gossip,
>> (does evolution have this input somewhere I can't remember?), and now
>> MC. I am not sure that using MC from the Nokia device is the best way
>> forward, but then I haven't looked at it enough yet to comment properly.
>  
> Gossip then, perhaps unfortunately, perhaps fortunately, would have the
> UI to do that manipulation and instruct the MC to log on and off. The UI
> that Gossip already has would be nice for this.

I think, for now, I would like to see account management done by Gossip
or gnome-about-me. I am not sure where this truly belongs. But again, I
need to look at MC to see what it does before really commenting on it.

>
> All of which is a lot of work.
>
> Good idea, bad idea? Got me. I do suspect that if Gossip doesn't do it,
> something else eventually will. I think that's basically just the
> direction our desktop (and other desktops) are heading in. I also think
> Gossip has a bunch of this work already done, and it would benefit us
> all of it was used instead of reinventing it again.

Yes, this is the nature of software. The same could be true of GTK+ if
more resources aren't put into it - resources are seriously lacking
here. Of course, this is an extreme, I don't mean to alarm anyone.

--
Regards,
Martyn
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

On dim, 2007-03-11 at 16:14 +0000, Martyn Russell wrote:

> Xavier Claessens wrote:
> > On dim, 2007-03-11 at 11:34 +0100, Tomasz Sterna wrote:
> >> I'm no big fan of telepathy. It's nice, yes.
> >> But it also bounds us. We cannot do and experiment with things that
> >> telepathy does not handle (yet). It also depends us hardly on something
> >> we don't fully control.
> >
> > We depend on Gtk, should be recode it to be sure to fully control it?
>
> That's a really bad comparison. What Tomasz is saying, is basically he
> can take Gossip and install it stand alone on his machine - he KNOWS he
> will need dependencies like GTK+ etc, but that comes for free in 99.99%
> of the cases, having Telepathy doesn't and that is quite important to
> remember here. If Telepathy was part of Gnome and could easily be used
> on other Operating Systems, then I think it would make the transition
> and idea more comfortable for everyone. At the end of the day users want
> to just install one thing, not a list of things. This is something the
> developer/maintainer has to consider in such situations.

Users don't care about installing application, distributions do.
Telepathy won't be installed by default if it's not used by the desktop,
and you say desktop won't use if it's not installed by default...

I think telepathy will be installed in all systems if we use it.

> > Do
> > one think but do it well, gossip is an IM client not an IM protocol
> > backend. I totally convinced that telepathy is THE solution for all IM
> > UI.
>
> Actually, that is wrong. Currently Gossip is an IM client with an IM
> backend and support for integration with an IM subsystem.
>
> >> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
> >> nicest. Dropping backend support in favor of telepathy would make it
> >> just another telepathy frontend.
> >
> > No, gossip is yet-another-jabber-client with nothing really special,
> > moving to telepathy opens the door to a new world of desktop integration
> > and would make gossip really interesting.
>
> I'm sorry Xavier, but I have to agree with Mikael here. That is a bit of
> a kick in the bollocks. If you feel that way about Gossip, why work on
> it? Why work on it in the first place if it is the same as all the rest?
> I am sure you don't agree with that comment especially with all the hard
> work you have put in.

Gossip is a jabber client like others. It has a better UI/code
quality/design/fun than all others clients, but it's still just a jabber
client... With telepathy and MC we can do much more and integrate IM in
the desktop.

> > For me it's simple, move to telepathy or die. Traditional clients like
> > gaim which re-implement all protocols has no future, it can't be easily
> > integrated with other desktop applications and has too many code
> > duplication with other traditional IM clients.
>
> Really it doesn't matter what GAIM does. I consider it far to broken for
> me to even worry about. But that is my personal view, I wouldn't be so
> hasty to condemn GAIM, it is still used by a LOT of people as Mikael says.
>
> > Take a simple example in gossip the telepathy backend supporting VoIP
> > and all possible IM protocols with almost the same features as the
> > jabber backend takes 6923 lines of code, gossip's jabber backend that
> > can't do msn/irc/etc and without VoIP takes 13691 lines of code... Using
> > telepathy we support far more with the half of code, that's the proof
> > there is a lot of code duplication in traditional IM clients.
>
> You are comparing apples to pairs here. I mean really, think about it,
> the Jabber backend has to interface to Loudmouth (which Telepathy does).
> So you are one layer higher with the Telepathy backend in Gossip. All
> the hard work by Daf, Robert, et al is done for you, all you have to do
> is connect a few signals, send a bit of data, remember some information,
> bla bla, it is a lot easier trust me.

yes, and all that work done by collabora guys is duplicated in
gossip/gaim/gajim/etc. If all IM clients depends on telepathy it will
save lots of work.

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Mikael Hallendal]

Xavier Claessens skrev:

Hi,

> For example gossip is (will be) the first IM client to support VoIP
> with jabber/gtalk thanks to telepathy. With that kind of think we are
> no more on the state of a simple jabber client, we do more!

Not sure what you mean here. There are several IM clients that support
voip. GTalk and IChat are two. Maybe you mean for GNOME?

I'm not convinced that putting voip support _into_ gossip is a good
idea. However, _integrating_ it with Gossip is. Be it through some
telepathy voip client or through Ekiga.

>> A nice UI is in my opinion what was lacking (and still to large
>> extent is) in other Jabber clients, so in that sense Gossip is not
>> "yet-another-jabber client with nothing really special".
>
> Totally agree, the UI is very nice! But I'm not talking about UI, I
> talk about the backend. Otherwise I wouldn't work to implement
> telepathy into gossip but I would start a new program designed from
> the ground to work with telepathy. I want to keep the ui and work
> only on the backend.

And here is where you are wrong in my opinion. Software design should
never be done from the backend up. Telepathy integration is not a goal
in itself. Design what you want in the UI-layer and then decide the best
way to do it technically.

Before you do this you can't really make any sound calls to the best way
to integrate it with Gossip.

> Users don't care about installing application, distributions do.
> Telepathy won't be installed by default if it's not used by the
> desktop, and you say desktop won't use if it's not installed by
> default...
>
> I think telepathy will be installed in all systems if we use it.

Yes, I get the feeling this is a lot of the driving force behind the
integration with Gossip. Ie. to drive Telepathy integration through a
popular UI.

> Dependencies are made to be used and to avoid duplicate work. That's
> true for gtk (to build UI) and for telepathy (to build IM)!

I do know why you put things in libraries yes. However, you are
comparing Gtk+ (over a decade years on it's neck) and Telepathy (don't
even have everything we need yet).

> True, gossip don't provide something better, but a telepathy/MC aware
> IM client with a good desktop integration does provides something
> totally new and far better.

Design this totally new and far better system/desktop and then we can
look into whether Gossip is the right way for this and if so, how to
change Gossip to fit the picture.

I'm not very interested in taking Gossip down a path of regressions and
unstability to try something out. Gossip is to mature for that.

Also, as I've said numerous times. Gossip does provide something better,
for Jabber users who want to text chat, keep that in mind.

>> Maybe you should attack the problem from the other way. Take parts
>> from Gossip and create separate small apps?
>
> I'm not sure for the best way here. First of all I really don't want
> to fork gossip UI and most of libgossip, if we split gossip into
> pieces I want to be sure people won't keep old gossip and continue
> work on it.

Why? Duplication of effort?

At the end of the day, people hack for different reasons. If someone
finds it more fun to hack on the old Gossip code base, let them.

>> My interest in Telepathy has nothing to do with multi protocol
>> support but the possibility to make XMPP more integrated in the
>> desktop. Whether this is done through Telepathy or a desktop
>> XMPP-server matters little to me (from a user point of view).
>
> Telepathy provides both advantages: desktop integration and multi
> protocol support. For me desktop integration is the best part and can
> begin if we adapt gossip to work with the MC.

Telepathy provides _a solution_ for desktop integration. However, at the
moment there isn't any and there also are other solutions for desktop
integration.

> yes, and all that work done by collabora guys is duplicated in
> gossip/gaim/gajim/etc. If all IM clients depends on telepathy it will
> save lots of work.

Rather, all the work spent in Gossip/Gaim/Gajim has been duplicated by
the Telepathy project. Telepathy is the new player here.

You are pushing new technology and trying to convince existing projects
to drop their tested code and move to something new and untested. It is
_never_ a good idea to do so without considering why things are as they
are now and what user bases Gossip in this case has at the moment.

Regards,
  Mikael Hallendal

--
Imendio AB, http://www.imendio.com/
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Martyn Russell-3]

Xavier Claessens wrote:

> On dim, 2007-03-11 at 16:14 +0000, Martyn Russell wrote:
>> Xavier Claessens wrote:
>>> On dim, 2007-03-11 at 11:34 +0100, Tomasz Sterna wrote:
>>>> I'm no big fan of telepathy. It's nice, yes.
>>>> But it also bounds us. We cannot do and experiment with things that
>>>> telepathy does not handle (yet). It also depends us hardly on something
>>>> we don't fully control.
>>> We depend on Gtk, should be recode it to be sure to fully control it?
>> That's a really bad comparison. What Tomasz is saying, is basically he
>> can take Gossip and install it stand alone on his machine - he KNOWS he
>> will need dependencies like GTK+ etc, but that comes for free in 99.99%
>> of the cases, having Telepathy doesn't and that is quite important to
>> remember here. If Telepathy was part of Gnome and could easily be used
>> on other Operating Systems, then I think it would make the transition
>> and idea more comfortable for everyone. At the end of the day users want
>> to just install one thing, not a list of things. This is something the
>> developer/maintainer has to consider in such situations.
>
> Users don't care about installing application, distributions do.

Actually, both do. Distributions care, but are willing to put a lot of
effort into getting packages working. Users expect things to just work
when they install them or they give up - and I know for sure that if
they find they have to install another 10 packages for a simple Jabber
client, they will be put off in some cases. Oh and user's DO care about
installing applications, because when they do do it, even if it is a
rare occasion, they want it to be fast, error free and to just work (TM).

> Telepathy won't be installed by default if it's not used by the desktop,
> and you say desktop won't use if it's not installed by default...
>
> I think telepathy will be installed in all systems if we use it.
>
>>> Do
>>> one think but do it well, gossip is an IM client not an IM protocol
>>> backend. I totally convinced that telepathy is THE solution for all IM
>>> UI.
>> Actually, that is wrong. Currently Gossip is an IM client with an IM
>> backend and support for integration with an IM subsystem.
>>
>>>> Gossip is a nice Jabber (XMPP in future maybe) client. One of the
>>>> nicest. Dropping backend support in favor of telepathy would make it
>>>> just another telepathy frontend.
>>> No, gossip is yet-another-jabber-client with nothing really special,
>>> moving to telepathy opens the door to a new world of desktop integration
>>> and would make gossip really interesting.
>> I'm sorry Xavier, but I have to agree with Mikael here. That is a bit of
>> a kick in the bollocks. If you feel that way about Gossip, why work on
>> it? Why work on it in the first place if it is the same as all the rest?
>> I am sure you don't agree with that comment especially with all the hard
>> work you have put in.
>
> Gossip is a jabber client like others. It has a better UI/code
> quality/design/fun than all others clients, but it's still just a jabber
> client... With telepathy and MC we can do much more and integrate IM in
> the desktop.

Some people only want a Jabber client. I think you need to understand
that. Currently we have:

- Gossip + Tested and more functional backend.
- Gossip + New slightly tested and less functional backend.

If you are only interested in Jabber, this is a step backwards in some
people's minds.

> yes, and all that work done by collabora guys is duplicated in
> gossip/gaim/gajim/etc. If all IM clients depends on telepathy it will
> save lots of work.

Like Mikael said, it is more like Telepathy duplicates what other
clients do, not the other way round. If all clients use Telepathy it
saves work in the long term, but in the short term, it generates a LOT
more work.

--
Regards,
Martyn
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Jerome Haltom]

Though this may be a bit too obvious, perhaps somebody who CARES enough
about this issue, and cares enough to dedicate time to it, could
simply... well... do something.

If somebody wants to break Gossip apart to make it more fitting for the
desktop he envisions, just do it. Eventually, if he's right, he'll gain
everybody's support. Or he won't. Only time will tell.

I'm sure everybody will assist this persons efforts the best they can.

Personally I think the telepathy approach is the "right" long term
approach. I think having a bunch of really useful reusable widgets and
interconnected pieces for IM-ish stuff is the "right" approach. Would
the result of this effort even be called Gossip anymore? I have no idea.

But I hold no expectations that my words alone will convince anybody on
this board or any others that that is true. Only actions can do that.


_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Mikael Hallendal]

Martyn Russell skrev:

Hi,

> What we or I have been doing wrong all these years is basically
> abstracting the whole contacts/presence/etc libgossip layer to work with
> multiple protocols, Mikael originally started doing the abstraction to
> make the code simpler to maintain. I think it wasn't a good idea and it
> hasn't done much for us at all, most of that code will be redundant if
> we choose Telepathy and drop the Jabber backend, and likewise, if we
> decided to do the same but for Jabber. This is my fault, I think I
> thought I could do it nicely and have backends written to just plugin.

Libgossip needed to happen and we have a far better and more
maintainable code base as a result of that refactor. However, I kinda
regret to add the extra layer to support multiple protocol backends as
we never went ahead and implemented any other than the jabber one.

On the other hand it made it possible to hack on a Telepathy backend
without interfering too much with the rest of the code.

Cheers,
  Mikael Hallendal

--
Imendio AB, http://www.imendio.com/
_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev

[Keywan Najafi Tonekaboni]

Hello,

I try to follow the discussion. Maybe I mist something, but I want to
share my opinion from a user view.

* I use Gossip because of it's great, clean UI and was ready to pay for
this good UI with a lack of features like file transfer. I hope Gossip
will keep this clean interface and focus on be a good IM-app (instead
e.g. good IRC support or so)

* I looking forward for telepathy, because it has the potential to
deliver a good support for a lot of protocols and features, even though
I only use jabber and try to convince others to switch to jabber.

* I hope programmers of different apps can work together on perfect
protocol implementation and have more time for their UI. If a feature is
implemented in telepathy, wouldn't it be able in all apps?

I want so much useful features (mainly file transfer and VoIP) as
possible, but with a clean design. By all respect for Gaim, I would
prefer to use MSN (the few moments I need it) with the for me familiar
Gossip-Interface instead of the Gaim-Interface.

Maybe I missed something, but I don't think their must be a decision
between the now great Gossip and Telepathy. I think telepathy could be a
great enhancement for Gossip.


Regards,

Keywan


--
Keywan Najafi Tonekaboni
http://www.prometoys.net

people@world:/# apt-get --purge remove dominion
After unpacking world will be freed.
You are about to do something potentially beneficial
To continue type in the phrase 'Yes, do as We say!'

_______________________________________________
Gossip-dev mailing list
Gossip-dev@...
http://lists.imendio.com/mailman/listinfo/gossip-dev