[Development] QtRemoteObjects (was Replicant)

[Simon Hausmann]

On Tuesday 21. October 2014 12.14.35 Stottlemyer, Brett wrote:
> On Monday, October 20, 2014 6:45 AM Simon Hausmann wrote:
> >> I wasn't trying to suggest using protocol buffers as a wholesale
> >> replacement for MOC generated code.  I was thinking only of the piece
> >> for marshalling Properties and Signal arguments for QueuedConnection
> >> calls, and the return value.  Since the types are known at compile
> >> time (based on everything MOC already does), it should be possible to
> >> generate the protocol buffer definitions for those items, which would
> >> then be compiled with the rest of the MOC generated code.
> > 
> > Hmm, the problem I see is that the types are _not_ known at moc execution
> > time, only at C++ compile time.
> I'm not sure what you mean.  The MOC output includes lines like:
> case 3: _t->timeChanged2((*reinterpret_cast< QTime(*)>(_a[1]))); break;
> 
> The QTime is determined by MOC when MOC runs....

The QTime is literally the result of string copy&paste from the token in the 
header file into the moc generated C++ code. There's no type lookup, resolution 
or analysis involved :)


Simon