Re: extending Gdk::Region

[Christopher Harvey <chris basementcode com>]

You used the term "Value Semantics" a lot in the last e-mail. What are
value semantics?

Daniel Elstner wrote:
> Am Mittwoch, den 06.05.2009, 13:42 -0400 schrieb Christopher Harvey:
>   
>> I want to extend Gdk::Region. I noticed it didn't have a virtual
>> destructor. I REALLY need to be able to extend it, otherwise I'll end up
>> with random functions floating around all over the place.
>>     
>
> A virtual destructor probably won't be of much help, as Gdk::Region
> implements value semantics so you can't really add any data to it and
> use it polymorphically.
>
> If you only want encapsulation but don't like the idea of inheriting
> from a class that doesn't have a virtual destructor, you could instead
> make Gdk::Region a member of your custom class.
>
>   
>> Basically I want to create a region class that can build itself from
>> complex input, then be used as a normal region.
>>     
>
> I don't know what you are working on; however, from your other mail it
> sounds as though you are trying to reduce the amount of drawing done in
> your expose handler.  If so, it's probably a good to check whether
> complicated regions will actually improve matters on average.  With
> Cairo clipping, it can easily happen that figuring out the clipping
> region actually increases the overall processing time.  It depends on
> what and how much you are drawing, I guess.
>
>   
>> Can we make this class extendable? It seems like a nice feature for a
>> widget library especially.
>>     
>
> Adding a virtual destructor probably wouldn't be a big deal here (but
> not until the next ABI break!).  However, I'm not sure it will be of any
> use, given the value semantics.
>
> --Daniel
>
>
>
>