Re: Group rotation

FROM : Graham Cox
DATE : Thu May 08 03:06:17 2008

I don't think it makes the more classic approach wrong - it's just 
another way to crack that particular nut, with its own advantages (and 
probably disadvantages too, like more code, more dependency between 
classes, possibly lower performance, I don't know - but there's no 
"one true way" to do it).

But either way, grouping is an exercise in concatenating transforms, 
basically.

I'm intrigued by your allusions to shadow drawing, I've had a lot of 
problems with that too - it's a performance killer.

Regarding flowing text around objects, do you have a function for 
breaking an arbitrary shape into a series of line rectangles that 
you'd be prepared to share with the world?



G.


On 8 May 2008, at 3:31 am, Gordon Apple wrote:

>    This is a major advantage of separating the drawing (rendering) 
> code
> from the data model instead of having objects draw themselves (not to
> mention cross-platform).  Our renderer object is recursive (and 
> could be a
> shared object if not for the deficiencies in shadow rendering -- a 
> story for
> another time).  When we replace objects with a group object, all the
> transforms automatically work.  Grouping can be nested to any 
> level.  Since
> we only have one "Shape" type, groups can also take on object
> characteristics such as fills, content, text, etc.  This especially 
> works
> well for using group objects for sophisticated text containers where 
> text
> flows around contained objects.
>
>    I know that about every example of drawing has subclassed objects 
> that
> draw themselves.  IMHO, that is the wrong approach.  There are just 
> too many
> advantages to separating the object rendering and making it part of 
> the C or
> V part of MVC instead of the M.
>