This in practice means that any HBQT based apps would have to
link in the *all* QT classes in final executable. Thus causing
significant bloat.
It's better to simply have them separately, and it's also much
better fits to current code layout. It also makes it easier to
debug.
Not exactly.
The if statement will be like this
if object == QObject ( 97% of Qt, may be more )
else if object == QSize
else if object == QPoint
Only very few are left, may be 20 at the most, and
all are used extensively in hbxbp.lib.
I ran into difficulties when I tried to write for each
class. Actually I had started like that but ended with what
you see in the zipped code.
I still can't imagine what sort of difficulties
you can get this way, it's used like this is every
place inside Harbour. Also the logic stays the same,
the separate one is cleaner, the merged one is more
error prone and less generic.
Also notice that for 'type' solution, you'll need
to maintain an extra "holder" structure, allocate
memory for it each time and release it each time.
So it's also much less efficient.
Brgds,
Viktor
_______________________________________________
Harbour mailing list
[email protected]
http://lists.harbour-project.org/mailman/listinfo/harbour
