[glade--]Re: glade-- patches

Bryan W. Headley wrote:
> Christof Petig wrote:
> 
>> Bryan W. Headley wrote:
>>  > [...]
>> Thank you for your valuable patch - it's already in CVS.
>>
>>> + handling of button, togglebutton, radiobutton in a Toolbar
>>
>>
>>
>> I do not agree to duplicate that code again in a new file ... does it 
>> help to register  Gtk_Button as 'button'? Or are GtkButton and button 
>> that different?
> 
> 
> That's a good religious question.
> 
> The buttons certainly have different sizing semantics (see picture). And 

Oh, please do not gzip .jpgs or .pngs - they will not shrink at all. And 
it makes it more difficult to see them. The 34k patch had been a good 
candidate for zipping.

> a "regular" Button has relief semantics. Translatable tooltips, stock 
> icons seems to be shared. The icons are generally 24x24, but I have 
> given 32x32 icons to a toolbar button without changing the overall 
> appearance/resizing of the toolbar.
> 
> Damon told me that it's sort of a kludge, that the only way they know 
> which behavior to give the widget is through a "secret" widget name 
> ('button' versus 'GtkButton'). Why didn't they just call the thing a 
> GtkToolbarButton, or GtkToolbarToggleButton is beyond me.
> 
> The crime's already been committed, but I don't know if I want to be a 
> criminal, too :-)

I'd vote for splitting them once we see the difference in gtkmm code 
needed. Until then I'd vote for ignoring this C hack.

> 
>>
>> proposal: in button.cc: Gtk_Button::Gtk_Button(bool)
>>
>> - Writer["GtkButton"]=this;
>> + { Writer["GtkButton"]=this; Writer["button"]=this; }
>>
> 
> I'd rather the 'button' writer be a separate object that inherits its 
> shared functionality from GtkButton's writer. And there should be a 
> clearcut division between the FILES and the writers contained in them.

perhaps. but I'd rather prefer to have less files. There already are so 
many. And if you look for a specific widget use fgrep. I even thought 
about putting all the buttons in one file. (but with the same class 
structure)

> (Doesn't matter, placeholders in 2.0 never find their writer, probably 
> because they don't have a <widget name='a_name'> tag, and end up lost in 
> Widget's check4validity(). Makes me think the XML parser might want to 
> ignore <placeholder/>, except that's wrong, too. Never write for today; 
> write for tomorrow; there MAY eventually be a widget that uses the '/>' 
> semantics that we care about...)

const_contained_iterator ignores them, now.

> --oo--
> 
> This is all your fault! Since you are making me study the .glade files 
> too closely,  you get to tell me what to do with the <accessibility> 
> attributes... The Glade C generator does something with these things.
> 
> AtkRelationSet, AtkRelationType, AtkRelation, AtkObject...  Time for 
> libgailmm :-)

I'd rather vote to have the basics work first ;-).

    Christof

PS: If things are of public interest, please post to the list so that 
others can benefit from your questions/insights. I'll see it no single 
second later.