[Psi-devel] Re: 0.10 and Qt4 Conversion (independant)

[Kevin Smith]

On 30 Jun 2005, at 16:40, Alex Mauer wrote:
> Remko Troncon wrote:
>>> I think we should do a MUC release before the Qt4 release.  
>>> Converting to
>>> Qt4 is going to be non-trivial and could take a while.
>> I suspect that integrating MUC will be a lot harder than porting  
>> to Qt4.
>> The Qt4 port should normally appear shortly after 0.10 is released.
> Hmm, I agree with Sneakin, but then I *really* want mu-c sooner rather
> than later.
Well, the way I currently see it is this: porting to Qt4 using  
compatability classes should be a matter of a few weeks if we're  
reasonably dedicated about it. It'll take a long time, probably also  
weeks, to review the muc patch for inclusion, less if nolan's been  
religious in following  our request to not touch anything unecessary  
in the patch; more if he hasn't. The testing period for such a large  
external patch has then got to be somewhere in the period of months.  
If we apply muc before qt4 then I'll have to port muc, if we do it  
after nolan can. As it's Nolan's code, the person best able to port  
it is obviously Nolan. Nolan could port MUC at the same time as we  
port Psi. So with all this considered. If we apply MUc before the  
port, we delay the port (and all subsequent features as there's  
little point devving for qt3 now) by several months. If we port  
before applying muc, we delay muc by a few weeks.

> mu-c has been working wonderfully for me. The only missing feature  
> (that
> I notice) is bookmarking conferences to the roster, and that's hardly
> critical.
As far as I know, nolan's muc code still causes segfaults on osx. The  
bookmarks in roster actually require the roster rewrite, which in  
turn depends on us porting to qt4.

> And I fear that porting to qt4, let alone distros actually having it
> available, will turn out to be more complex and/or take longer than
> expected.
On the one hand, users with particularly out of date linux distros  
will be unable to run the new Psi without compiling Qt. Mac users  
will be unaffected; they either run the builds, or compile qt  
themselves. On the other side, suddenly windows devs will be able to  
develop without the Qt license. Hoorah.

I've considered it in reasonable depth now, and have made reasonable  
effort to ascertain how hard the Qt4 port will be, and I think this  
is the best plan for moving Psi forward.

/K

--
Kevin Smith
Psi Jabber client maintainer (http://psi.affinix.com/)
TaeKwon-Do Club Captain (outgoing) University of Exeter
Postgraduate Research Student, Computer Science, University Of Exeter