yang686526

what is 'pendpointsoverrides'
what is 'pendpointsoverrides'
please,
what does the 'pendpointsoverrides' parameter in odgigeometry::elliparc() do?
thanks
arnošt

these parameters specify start and end point overrides for elliptic arc. to understand their necessity, imagine situation:
you have ellipic arc a with start point a and end point b.
you need to transform it by matrix m.
you will get arc a*m. you expect that its start point will be a*m. but it is not always true - due to loss of precision (a*m).startpoint() may be a bit different with a*m. overrides are used to render start and end points precisely.
sincerely yours,
george udov
thanks george
actually...,
as i am further thinking about the correct implementation, the overridden points do not make sense if either the arc is circular or closed or both. am i right? the way i see it is that if an arc is closed there are no endpoints. and if an elliptical arc is de-facto a circular arc then i would not know how to figure its center. should i use the arc's center (arc.center) and modify it by the difference between the arc's and points and the overridden points? does it make sense? of course, if it does, both differences must be the same.
i think it's a bit confusing. what if one wants to tessellate such an arc. he/she can set the precise end points but what about the mid points. can arc.evalpoint() be used? should a correction be made to those points as well? and what about the other arc's methods, such as majoraxis? does the returned vector have to be modified or is it correct regardless of the overridden end points?
arnošt

let's return to our sample.
you have ellipic arc a with start point a and end point b. this arc has center c, major axis d, minor axis e, start parameter f, end parameter g. c, d, e, f, g are parameters that stored for arc (odgeelliparc3d which is used in odgiconveyorgeometry::elliparcproc). a and b are not stored in odgeelliparc3d.
you need to transform it by matrix m.
obviously,
(a*m).center() == c*m
(a*m).majoraxis() == d*m
(a*m).minoraxis() == e*m.
this is obviously due to algorithm of odgeelliparc3d::transformby().
but a and b are not stored. so when we obtain a.startpoint() and a.endpoint(), they are calculated. start and end parameters (which are stored) are transformed in more complex manner. first, start and end points are calculated. then they are transformed. then parameters are calculated. when we take this parameters and calculate points next time, points will different with ones from which we obtained start and end parameters, due to approximate calculation of trigonometric functions by pc. of course, difference will be very little, and in many cases it can't be ever noticed. but there are cases when this difference can influence arc's apearance.
sincerely yours,
george udov