agg rendering offset x -99 lines trouble

[szigeti]

There are adjacent lines of direction dx/dy==2 (probably also 1/2, -1/2 or -2) in the same shape object. E.g.: FEATURE POINTS 0 0 200 100 300 150 END END . Offset style is used with y=-99. This scenario may turn the agg renderer into long (probably infinite) loop.

The attached mapfile contain various examples (commented out) related to this bug.

[szigeti]

My home-made bugfix: set float par=0.7071 instead of 0.71 for better sqrt(0.5) approximation in shapeObj* msOffsetPolyline(shapeObj *p, double offsetx, double offsety) function (maputil.c).

Note, it is not the best overall solution, but for me it works:)

[szigeti]

rewritten msOffsetPolyline that fixes the bug

[szigeti]

​http://trac.osgeo.org/mapserver/attachment/ticket/2868/maputils.patch

I have introduced some vector primitives (pointObj functions), e.g., addition subtraction, resize, rotation, etc. and have rewritten the msOffsetPolyline function. Feel free to to apply this changes to the trunk in your own coding style.

Note that for the sake of simplicity I have used the type pointObj everywhere insead vectors - in fact point - point = vector, point + vector = point, vector + vector = vector, etc.

[sdlime]

Seems to work ok, want to look things over before committing...

Steve

[sdlime]

The limited testing I did was successful (perhaps GD should use this too). I did just a bit of format clean up and committed in time for beta 2 (r8613). Thanks for the patch!

Can I assume all contributions are legally unencumbered (i.e. yours to make) and are now covered by the MapServer license?

Steve

[szigeti]

Replying to sdlime:

Can I assume all contributions are legally unencumbered (i.e. yours to make) and are now covered by the MapServer license?

Yes, of course, and moving the vector helper functions into mapprimitive.c is also a good idea. But these vector primitives are certainly implemented in other libs as well. Just to avoid redundancy:)