RE: [gtkmm] Gtkmm or Qt?

[Dr Mark H Phillips <mark austrics com au>]

On Tue, 2003-08-26 at 17:39, Murray Cumming Comneon com wrote:
> I believe he is a fool. Large parts of that article actually agree with me.
> Most imporantly, he has almost no knowledge of gtkmm 2. But you can make up
> your own mind.
> 
> All of this has been discussed thoroughly all over the net. The thread here
> is fairly representative:
> http://linuxtoday.com/news_story.php3?ltsn=2003-04-01-032-26-OP-DT-DV&tbovrm
> ode=3#talkback_area

Thanks for this link.  It gives me a bit better idea of the issues. 
There seem to be lots of minor ones, but two issues that seem more
significant to me are:

1. It would appear that Qt handles internationalization support better.
Now probably this would be better done as a separate library??  Does
one exist or is one planned?

2. Laurent seems to think that the signals and slots mechanism provided
by MOC gives capability that the gtkmm library, libsigc++, doesn't or
can't.  I don't know enough about these issues to be sure about who is
right.  Certainly Trolltech think a template solution will never be
as good.  On their web page:

  http://doc.trolltech.com/3.2/templates.html

they outline there reasons why they would still use their MOC approach,
even if properly working template C++ compilers were guaranteed always
to be used.  Their key arguments seem to be in the section "flexibility
is king".  They talk about the ability with MOC, to overload signals and
slots, to "have object properties", to "add new signals without breaking
binary compatibility" and to "explore an object's signals and slots at
runtime".  Perhaps this is what Laurent means when he complains about
libsigc++ lacking introspection??  Trolltech seem to be saying that
runtime introspection allows GUIs to be generated and connected
on-the-fly in a way which is impossible with a template slots and
signals library.  But is this correct?  Isn't this what libglademm
allows you to do?  Or am I misunderstanding?  Do you think Trolltech's
comments about the superiority of their system over template based
approaches refers to older approaches and don't apply to libsigc++?
Or is there some degree of validity to their points?

Thanks,

Mark.