Re: [Gnash-commit] [SCM] Gnash branch, master, updated. release_0_8_9_final-2083-g223b921

[Bastiaan Jacques]

On Fri, 30 May 2014, Richard Wilbur wrote:

> On Fri, May 30, 2014 at 6:30 AM, Bastiaan Jacques <address@hidden> wrote:
> > stored. It would probably be more elegant to store the control point
> > only for actual curves; patches are welcome.
>
> So (dx, dy) are not differentials in x and y but the coordinates of
> the new point?  If they were differentials I would expect to store
> something like
>      new_x = dx + current_x;
>      new_y = dy + current_y;
>      m_edges.emplace_back(new_x, new_y, new_x, new_y);

No, these are absolute coordinates. (Actually, SWF files provide
relative coordinates, but they are converted to absolute coordinates by
class ShapeRecord.)

> You mention an Edge as storing the curve's control point.  Is that in
> addition to the end point?  If so, is that the way we tell that this
> is a straight line instead of an actual curve?

The way we can tell this is a straight Edge (line) is by checking if the
control point is the same as the destination point, which is just what
Edge::straight() does.

But it would be better if we had structs LineTo { point dest; };
CurveTo { point control; point anchor; }; MoveTo { point dest; }; and
then Path could have a vector<boost::variant<MoveTo, LineTo, CurveTo>>
instead of vector<Edge>. Or something along those lines.

Bastiaan