Re[2]: [Gnash-dev] Coordinate system discussion cont'd

[Udo Giacomozzi]

Hello Bastiaan,

Wednesday, May 7, 2008, 6:44:44 PM, you wrote:
BJ> Actually, I think we should take the opportunity to resolve some
BJ> ambiguity with regards to the word "world". In the renderer, I believe
BJ> it's used to convert various coordinate systems (e.g. SWF def.) to renderer
BJ> coordinates and vice versa.

You mean AGG renderer, correct? Right, there's a world_to_pixel and
pixel_to_world. I got used to these terms because I always has such
functions in my graphics projects.
Anyway, these functions will probably become (or already are, since
AGG already does) useless when the renderer uses a matrix.

BJ> So (I think) world here means coordinates in
BJ> TWIPS, with renderer matrix applied.

It's basically used to apply scaling in AGG.

BJ> However, in libgeometry, I believe
BJ> world means something different namely, in a Range2d world means "contains
BJ> all possible points".

I'd say "extended to infinity in all directions".

BJ> In the character class, there are the get_world_{matrix,cxform}
BJ> methods and I don't know what world means in that context.

All matrices concatenated, that is, the movieclip's matrix
concatenated to all parents up to _root. So you only have to apply
one matrix to a point to get the effective coordinate.

It's used in lots of places and changing it's name would probably
confuse us even more. I think in this regard "world" is not a bad
choice. Also, there is not a high risk to confuse it Range2d's
setWorld().


BJ> So I'd definitely like to either clearly document the differences, or,
BJ> even better, use world to describe only a single thing.

I agree that we should remove (or rename) world_to_pixel and viceversa.

Udo