[Interest] can the QRegExpEngine cache acces objects belonging to an unloaded library during global destruction?

[René J. V. Bertin]

Thiago Macieira wrote:

> Correct, you don't need the macro for the new-style connect, except to add new
> signals. But add it anyway.

That means I'll need a moc file. Is that going to oblige me to move the class 
definition to a header file, or is there another way to obtain that moc file?

>> - (ahem) do I need to call the parent default constructor explicitly if I
>> override (extend) the default constructor?
> 
> If you're passing any arguments.

Yay for consistency!

> Your answer is in the question itself: if a thread is doing something, can it
> at the same time be doing something else? When you say "non-deadlocking lock",
> you mean "reentrant mutex".

Not at the same time in the absolute sense of the term, of course. But you're 
right, it seems unlikely a thread would send itself a signal from deep below the 
event loop. I guess I still confound Qt signals and system signals. 

Damn, in the meantime I've come up with a rather nice synchronisation mechanism 
(featuring a timed QSemaphore), now I can break all that down again :)

R.