[Development] [fix]: Assistant (and others) burning CPU because of #f27d1ccbb24ec2fd4098f2976503478831006cc8 ?

[René J. V. Bertin]

Eike Ziller wrote:

> Another good question to ask when introducing source incompatible changes, is
> “how hard does it make supporting building with both Qt 5 and Qt 6 from the
> same code base”. We did that for Qt Creator for quite some while, and it makes
> the transition process much smoother than doing a big jump with no way back.

Agreed. You also risk ending up with a growing amount of "junk DNA" but I concur 
that's a lesser issue. (Thinking of that other framework known with the same 2-
letter acronym which in the end was replaced by a completely new and clean 
"slate").

> Not sure what kind of hacks could be done for that specific case (maybe adding
> a funny custom QObjectTimer class for the Qt 5 build).

That was my thought too. If Qt6 moves startTimer() and killTimer() to a 
dedicated QObjectTimer class, code wanting to support Qt5 can simply introduce a 
bogus QObjectTimer class. (The Qt5 LTS releases might even do that too.)

Anyway, I drop my idea of adding some kind of protection to QObject's 
timerEvent() method. I added a qWarning() probe to the method and realised that 
evidently I had forgotten that one can also handle timer events in a 
QObject::event() override...

R.