enable-final and QT_DECLARE_METATYPE

[Stephen Kelly]

Will Stephenson wrote:

> On Monday 06 December 2010 22:06:05 you wrote:
>> On Monday 06 December 2010, Will Stephenson wrote:
>> > There have been a couple of enable-final breakages while preparing
>> > 4.6beta1 and now beta2 tarballs that keep coming up, so I'd like to
>> > remind you about enable-final (now in the KDE 4 epoch
>> > -DKDE4_ENABLE_FINAL).
>> 
>> Actually just last week I wondered whether this is still used by anybody,
>> and whether it is still worth to keep it working now in the days of
>> multicore. So, is it ?
> 
> We use it for openSUSE builds.  I haven't benchmarked it.  A priori it
> should save IO time and cache though.

I'd also be interested to know if KDE4_ENABLE_FINAL is worth it. Getting rid 
of it would mean one less reason for kde4_add_{library,executable} to exist. 
I'd always wondered why we needed our own prefixed macro, and can't just use 
the cmake one.

If it is worth it, can we try to get it (or something better) upstream to 
cmake. enable_final doesn't/can't cat the moc files together with the rest, 
whereas that could be done by cmake at the right time after the dependencies 
have been worked out.

> 
> Will