Re: gtkmm-2.4/MSVC in CVS

[Cedric Gustin <cedric gustin swing be>]

Hi Timothy,

Timothy M. Shead wrote:
> ... works fine, except for our One Big Ongoing Problem.  I've submitted 
> a patch to get the remaining demos to build, nothing fancy was required.

Interestingly, I had just commited the same changes to CVS ! Sorry for 
the duplicate work. The only difference is the get_value for the slot in 
demowindow.cc: I use "const type_slotDo& slot = row[columns.slot];" 
while you prefer "type_slotDo slot = row.get_value(columns.slot);". Both 
work of course but I was wondering which one was intrinsically better ?

> Amusingly, I can confirm that the RTTI problem persists whether using 
> GTK+ build with MinGW or MSVC.

Same here. It's too bad because apart from this RTTI bug, gtkmm works 
very well with Visual Studio.

> I haven't had any luck creating a "C++ only" test case yet, will 
> continue to plug away on it.

I strongly believe there must somewhere in the gtkmm code a dynamic_cast 
to a derived class that has not been fully created yet (as you pointed 
out in your MSDN link).

Murray, maybe you'll have some time this week-end to build glibmm/gtkmm 
with Visual Studio ?

Cedric