Re: [glade--]Re: glade-- and gtkmm-1.3

[Christof Petig <christof petig-baender de>]

Damon Chaplin wrote:
> On Fri, 2002-04-12 at 12:42, Christof Petig wrote:
> > Petr Ferschmann wrote:

> By they way, do you still use the extra C++ properties we added
> (Separate File, Use Heap, Separate Class, Visibility)?
> We added an agent parameter to the property element so we can still 
> include these in the XML, e.g.

Use Heap is no longer needed (since gtkmm-1.2). Separate File depends on 
Separate Class, so if you have an easy mechanism to code that an option 
is only sensible if another one is active, can you apply it here?
(Oops my English plays tricks on me: If it's easy for you to do it and 
you find time for it, please code it. But not having it doesn't hurt much).

>   <property agent="glademm" name="separate_file">True</property>
>
> Does that sound OK?
>
> Unfortunately I haven't updated Glade to load/save that yet, but I
> should get to it this week.

but <property name="cxx_separate_file">True</> still worked up to now, 
didn't it? I'm fine with both but marking this property as glademm 
specific is definitely a good idea.

----- I have some long lasting ideas for future expansions (simply 
didn't find the time to code it in glade) ------

In C++ a signal callback's data needs to be typed. Up to now I use a 
heuristic: 'true' or 'false' -> boolean, decimal number -> int, else 
string. But Sometimes widget pointer or enums make sense. IIRC glade 
always uses const char * cast to a gpointer?

- an additional data type entry for C++ (within Signals), in C++ even 
more than one parameter is possible (one of gtkmm's real strengths!)

-----

Conditional Widget inclusion would be another fine feature. Say you only 
want a widget if defined(EXTENSION_FOO) ... Doesn't that sound a nice 
feature even for C? But of course this feature (once simply implemented) 
would call for a lot of UI (show me the window with EXTENSION2 defined) 
to make it work smoothly.

   Christof