https://issues.apache.org/ooo/show_bug.cgi?id=115917

Armin Le Grand <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ACCEPTED
     Ever confirmed|0                           |1

--- Comment #4 from Armin Le Grand <[email protected]> ---
ALG: Good catch. impCorrectContinuity only corrects the numerical error in the
conversion from integer to double precision polygon, thus only errors in the
range of 0.0 <= error <= sqrt(2.0) can happen when the one vector is offset on
integer on +-1 in X and/or y.
Choosing the longer (originally integer) vector bets on the fact that it might
have been snapped better to integer than the shorter one. An even better
correction would be to take the mediated direction for C1 correction and the
mediated control points for C2 correction.
Checking how to do that...

-- 
You are receiving this mail because:
You are on the CC list for the bug.

Reply via email to