[GRASS-dev] Re: SoC Report: v.buffer/v.parallel reimplementation

[Росен Матев]

This week I finished the algorithmic part of buffer lines creation.
The main problem which ocured during tests is due to floating-point
accuracy errors. If I manually set the tolerance it works. The one
approach I tired to compute tolerance from the input map failed. I am
starting to think there might be a need for more precise calculations
at certain points of the code (more than double can give us).
Except this <little> issue, buffer lines are working fine. Next week,
I'll incorporate everything into v.buffer2. I will also do research on
floating point calculations and see what I can do with all the RETs
mess. I already have vague idea how to cope with the issue. I still
have issues with parallel lines. Yes, buffers are ok, but parallel
lines are not, because I don't use those parallel lines for buffer
creation. So defining parallel lines behavior is still open topic.
Probably I should write in ml for help with the last one.

Rosen Matev