Qt 3.2 requirement

[Cornelius Schumacher]

On Monday 28 July 2003 12:02, Chris Howells wrote:
> On Monday 28 July 2003 10:49, Chris Howells wrote:
> > > Naturally some people want to play with new stuff, that is perfectly
> > > fine. Others want to keep stuff stable, that's also a worthwhile  goal.
> >
> > We're talking about CVS HEAD here. HEAD is not stable. That's the point.
>
> Let me expand on that.
>
> KDE 3.2 _will_ depend on Qt 3.2 when KDE 3.2 is released (if it doesn't
> it'll be extremely disappointing for users of the new languages supported
> in Qt 3.2). I don't think anybody is in any doubt about that.

What changes in KDE are required to support the new languages?

> Switching to Qt 3.2 will allow us to test Qt 3.2 and find any problems
> leading to a nice stable KDE 3.2/Qt 3.2.x combination when KDE 3.2 is
> released.

Yes, we certainly should thoroughly test KDE 3.2 with Qt 3.2.

> Your argument doesn't make any sense at all to me. Putting off the
> inevitable to the last minute potentially means that the KDE 3.2/Qt 3.2
> combination will be _largely untested_ causing problems which will have us
> wishing we started testing with Qt 3.2 as a requirement sooner.

If there is a stable Qt 3.2, the KDE 3.2/Qt 3.2 combination will certainly not 
go "largely untested", as at least a part of the developers will upgrade Qt, 
even if it isn't required by KDE.

> In one of your previous mails you send "You unnecessarily force all the
> developers to test QT". Well if the developers don't test Qt, who do you
> think will? The users who have crashes and report them on http://bugs.kde.
> org?

It's good if developers test Qt. That's certainly in the interests of 
TrollTech and also in the interests of KDE if this leads to a more stable Qt 
and less problems, when Qt is upgraded.

The question is, what do we gain by introducing a requirement on Qt 3.2 and 
what do we loose by supporting both Qt 3.1 and Qt 3.2 for at least some time. 
If the only gain is that Qt 3.2 is tested I think it's not worth the trouble 
it causes for application developers who are forced to upgrade Qt without 
gaining anything.

It's a frustrating experience, if you want to quickly fix a bug and then 
realize that you first have to compile a new version of Qt, before you can 
test your fix. This costs us developers.

-- 
Cornelius Schumacher <schumacher at kde.org>