[Gossip-dev] Re: [Telepathy] Telepathy Status

[Martyn Russell]

Jimen Ching wrote:
> On Thu, 12 Oct 2006, Martyn Russell wrote:
>> As some indication of how long these things take, we (Micke and I) spent
>> the best part of 3 months refactoring the code to add libgossip and make
>> the middle-ware of Gossip generic enough for this sort of Technology
>> (Telepathy). I and others have also spent > 6 months writing the UI to
>> work with n+1 accounts in the chat rooms window, the accounts window,
>> the personal information window, the add contact window, etc. This is
>> just adding n+1 accounts, imagine how long it will take to add n+1
>> account types given that they also don't all operate in one way
>> (consider IRC and Jabber).
> 
> Hmm, this worries me some.  I might have not understood how Telepathy
> works, but I've always assume Telepathy handles these account issues
> already.

If you are talking about the refactoring, that is just an example of how
long it takes to write the software.

If you are talking about operating with n+1 account types, then NO
telepathy doesn't do the UI it just provides the communications mechanism.

>> What is the point in having a schedule if the work input is not
>> consistent enough to guarantee such a schedule could be kept to?
> 
> Having a schedule helps you track where you are and how far you have to
> go.  A schedule is just a plan, a todo list, etc.  I'm pretty sure you
> have such a schedule/plan/list.  It's probably only for the short term,
> and isn't written down.  I'm merely suggesting such a list be created,
> so everyone knows their part in reaching a goal.
> 
>> are really basic and a necessity for Gossip) and there will be bugs. If
>> you consider that it might take 3 months for Telepathy to get to a place
>> where they support all the key features of Gossip now and that it would
>> be stable by then too, that leaves the Gossip team (of 2/3 people plus 2
>> volunteers) 3 months to implement all the backend, all multi protocol UI
>> updates and any fixes in between. This leaves no time for any testing
>> what so ever.
> 
> If you're trying to convince me that there are lots of work to be done,
> you don't have to do that.  As I look through the code, I can see there
> are work to be done.  But the qoute above seems to imply things need to
> be done serially.  I.e. Telepathy needs to stabilize before work on
> Gossip can begin.  

Work has begun, on the TELEPATHY branch, but merging that work on to
HEAD before Telepathy is stable would be a mistake.

> I'm not sure this is true.  If work progress in
> parallel, based on features, it should be possible to reach the same
> goal at the same time.

In practice this never happens. Purely because we don't have enough
people working on Gossip and because you always have to wait for
Telepathy to do it, test it, release it, etc before you can use it. To
do so before it is released, wastes time which we don't have a lot of.

> I know this thread is taking time away from important bug fixing and
> feature enhancements.  But I get the feeling that a lack of a plan is
> what's preventing significant progress.  For example, Eitan did a lot of
> work to create a telepathy layer.  Xavier came into the picture and
> moved all of that into the protocol subdir.  Now, there's a suggestion
> of merging telepathy and libgossip (because they are basically doing the
> same task, abstracting the protocol).  Eitan had this in mind from the
> beginning.  So we're just moving back-and-forth.  Some planning could
> have prevented all this.

Nothing is going back and forth. The Telepathy implementation has always
been a protocol implementation in Gossip. All _I_ did was move the
directory to the protocols/ directory.

-- 
Regards,
Martyn