Re: [Lib2geom-devel] Dropping invalid arc segments in parser

[Krzysztof K. <twe...@gm...>]

2014-08-05 21:29 GMT+02:00 Johan Engelen <jbc...@sw...>:
> Ok, I see you're saying that "a 20 30 45 1 0 0,0" is ambiguous, hence
> invalid and the parser can skip it. (degenerate line segments are not
> ambiguous)
> This would vote in favor of floating point comparison, but I think in
> the interest of (human) predictability, the equals-check should be made
> are_near. Do you agree?

The epsilon of this comparison should not be hardcoded. For the case
of Inkscape, I think it should depend on the numeric precision of
output.

(PS I somehow missed this thread and spammed the Inkscape mailing list
about this.)

Regards, Krzysztof