Hello all,

Currently, LPEs have a 'doEffect' chain: basically a stack of functions that are called sequentially that convert from Inkscape's path type, to several 2geom types in steps. An LPE defines a doEffect within this chain to override the default behavior (pass on to next down the stack). The problem is this: Inkscape's paths can be discontinuous, meaning for the user that visually they are 2 separate paths. The LPE doeffect chain now converts it to one path. This means that for example lpe-PatternAlongPath will see discontinuous paths as a single path (thanks JFB!), so when applying a pattern (copy modus: single stretched) the pattern is applied along the whole path, ignoring the discontinuity. 1. Take two separate paths and combine them into one (ctrl+k) 2. draw a circle, convert to path, copy it 3. apply lpe-Pattern Along Path to (1) and paste the circle as pattern 4. the result will be two circle halves. where actually I would have expected to see two circles, since it looks like the original path is 2 paths.

This is undesired behavior I think. Note that e.g. lpe-Gear also suffers from this: it just seems to ignore the second path.

Do people consider this as a bug, or as a feature? The solution is to break the path into separate parts, group it, and apply an LPE on the group... but... especially with the lpe stacking, this is no longer possible and we do need to consider what lpe should do with discontinuous paths. I have made a fix, which I will probably apply to the whole doEffect chain. But perhaps people do not see this as a bug in PatternAlongPath, which means I will have to add a checkbox for the new behavior.... Basically, instead of concatinating all paths, run the effect multiple times on the differnt paths:

Technical, instead of:

std::vector<Path> doEffect_path (std::vector<Path> & path_in) { Piecewise<D2<SBasis> > pwd2_in; for (unsigned int i=0; i < path_in.size(); i++) { pwd2_in.concat( path_in[i].toPwSb() ); } Piecewise<Geom::D2Geom::SBasis > pwd2_out = doEffect_pwd2(pwd2_in); std::vector<Path> path_out = path_from_piecewise( pwd2_out, LPE_CONVERSION_TOLERANCE); return path_out; }

I want to change it to:

std::vector<Path> doEffect_path (std::vector<Path> & path_in) { std::vector<Path> path_out; for (unsigned int i=0; i < path_in.size(); i++) { Piecewise<Geom::D2Geom::SBasis > pwd2_in = path_in[i].toPwSb(); Piecewise<Geom::D2Geom::SBasis > pwd2_out = doEffect_pwd2(pwd2_in); std::vector<Path> path = path_from_piecewise( pwd2_out, LPE_CONVERSION_TOLERANCE); for (unsigned int j=0; j < path.size(); j++) { path_out.push_back(path[j]); } } return path_out; }

Note that: Piecewise<D2<SBasis> > doEffect_pwd2 (Piecewise<D2<SBasis> > & pwd2_in) ==> this one does the actual effect

Thanks for your consideration. I'll make sure this will not break the other effects.

Regards, Johan