Re: Overriding property defaults

[Owen Taylor <otaylor redhat com>]

Hey Tim,

Did you have any thoughts on this? 

 - Whether we should be worrying about this problem
 - Whether the approach is good

I'm going through the open GObject bugs, and trying to decide
whether to punt this one (#51748) from the 2.0 milestone.

The minimum to be able to do this without breaking source
compatibility is adding _copy and _set_default functions to
GParamSpecTypeInfo.

Regards,
                                        Owen

Owen Taylor <otaylor redhat com> writes:

> While reviewing at a patch to clean up GtkButtonBox, I noticed a
> problem with the way properties work currently.
> 
> The problem is that GtkButtonBox inherits GtkButtons' spacing
> parameter, but wants to change the default value. It is currently
> impossible for a derived class to modify the default value for a
> property.
> 
> I think this is a fairly common requirement. (Somewhat less common,
> but also legitimate, would be wanting to use more restrictive
> validation for a property.)
> 
> The essence of my proposal would be:
> 
>   GParamSpec *spec = g_object_class_copy_parent_property (object_class, "spacing");
> 
>   g_param_spec_int_set_default (G_PARAM_SPEC_INT (spec), BUTTON_BOX_SPACING);
>  
>   g_object_class_install_property (object_class, 0, spec);
> 
> That is:
> 
>  - We add a way of copying paramspecs, and a convenience function
>    in GObject to copy the parameter spec of a property you are overriding.
> 
>  - We add a way of setting default values; another possibility would be
>    to have a virtual function in GParamSpec taking a GValue, so you could
>    do:
> 
>     g_param_spec_set_default (spec, BUTTON_BOX_SPACING);
> 
>  - We allow properties to be installed with a parameter ID of 0, meaning
>    "ignore this property when looking for the getter/setter".
> 
> This requires some changes to the way the internals of GParam works,
> in particular GParamSpecPool. And for efficiency/simplicity one might
> want to move the ParamID directly into the GParamSpec instance as a
> private member.
> 
> But I think, in general, it is a fairly small change and a quite
> important feature we are currently lacking.
> 
> Regards,
>                                         Owen
> 
> _______________________________________________
> gtk-devel-list mailing list
> gtk-devel-list gnome org
> http://mail.gnome.org/mailman/listinfo/gtk-devel-list