【转帖】incorrect mirroring on 2d polylines
 

incorrect mirroring on 2d polylines ?
incorrect mirroring on 2d polylines ?
in the attached file there are numerous 2d polylines that are mirrored about the x and/or y axis. my application is creating the transformation matrix, applying it to the entity, and then exploding the result into the individual entities. it seems that after the transformation and mirroring about an axis the arc entities within the polyline are not being reversed. all the arc normals remain at 0,0,1. the code below illustrates the issue: i am using version 2.01
static void debug11314 (oddbdatabaseptr pdb)
{
oddbentityptr pentinsert, pent, pentcopy;
v_entity_type entitytype;
odgematrix3d odobjectmatrix;
odresult odret;
oddb::poly2dtype polytype;
//blockref
pentinsert = pdb->getoddbobjectid(oddbhandle("4f697")).openobject() ;
//get the transformation matrix
if (getobjecttransform (pentinsert, entitytype, &odobjectmatrix) == false)
return;
//entity
pent = pdb->getoddbobjectid(oddbhandle("4f5d0")).openobject() ;
odret = pent->gettransformedcopy (odobjectmatrix, pentcopy);
// print out arc data for original & transformed entity
debug11314readbyexplosion (pent);
debug11314readbyexplosion (pentcopy);
return;
}
static void debug11314readbyexplosion (oddbentityptr pent)
{
odrxobjectptrarray entityset;
odresult odret;
odret = ((oddb2dpolylineptr)pent)->explode (entityset);
noentries = entityset.length ();
if (noentries == 0)
return;
odrxobjectptrarray::iterator pit = entityset.begin();
for(; pit != entityset.end(); ++pit)
{
pobj = pit->get();
if (pobj->iskindof(oddbarc::desc()) == false)
continue;
parc = pobj;
as = parc->startangle ();
af = parc->endangle ();
debug_printf ("angles = %f, %f \n", as, af);
normal = parc->normal();
debug_printf ("normal = %f, %f, %f \n", normal.x, normal.y, normal.z);
}
}
regards
gerry
as i am unable to upload file i will mail separately
mirroring transformation about x or y does not reverse normal.
explode produces arcs having the same normal as original polyline entity. this causes direction of clock-wise segents to be changed (start and end points are exchanged).
this is intended behavior (like autocad does).
sergey slezkin
thank you. i did not realise that mirroring retained the same normal, and in effect reversed the polyline. nevertheless in this example the mirroring on arcs does seem to be incorrect. to simplify the data i have used the same transformation matrix (handle 4f697), but set the translation co-ordinates to zero, using the statements
n.x = 0.0; n.y = n.z = 0.0;
odobjectmatrix.settranslation (n);
the matrix now represents a mirroring about the x axis. if i explode the polyline (handle 4f5d0) and examine the 2nd line and 5th arc (because they are physically close to each other) of the original and transformed polylines i get the following results. all z co-ordinates are 0.0
for the 2nd line from the original polyline
start x-12.711 y35.094
end x-44.526 y0.387
for the 2nd line from the transformed polyline
start x-12.711 y-35.094
end x-44.526 y-0.387
these results show a mirroring around the x axis.
for the 5th arc from the original polyline
centre x-55.055 y-0.000000 rad 9.0 (y is actually -6e-11)
start x-58.797 y8.185
end x-46.576 y-3.018
normal x0.000000 y0.000000 z1.000000
for the 5th arc from the transformed polyline
centre x-50.318 y-5.167 rad 9.0
start x-58.797 y-8.185
end x-46.576 y3.018
normal x0.000000 y0.000000 z1.000000
clearly not mirrored about the axis. the centre co-ordinates should not change in x, and if the centre y co-ordinate is 0.0, then the mirrored object co-ordinate should also be zero.
regards
gerry
i reproduced the problem. the defect is in oddb2dpolyline::transformby(). it should negate bulge value if tranformation is mirroring.
sergey slezkin
is this a bug that is going to be fixed?
is this a bug that is going to be fixed?
what is the best workaround?
this bug was fixed in 2.3.1
sergey slezkin