[fixed] New MMB Connect for vertices (BUG)

[ggaliens]

Manifoldlab version that I based pull request on will do this configuration just fine. Truth be told ... after a day full of testing. I prefer my own version for many reasons. But I will support Dan's versions as time permits. I really really don't want to have two versions around too. So we must continue to collaborate.

[dgud]

"Truth be told ... after a day full of testing. I prefer my own version for many reasons",

Nice, there was a reason I replaced your version and still gave you credit, but I do not want
to thrash people on the internet.

[Dimitri]

It is best to have both of the two versions. Why not? It is good to have many ways of achieving a result. And if something does not work in one of the two versions someone may turn to the other version. Richness of choices! : - )

I have played a lot with the command in both of the two versions (the official one and the ML one)... in both of them there appear -in some cases- some unpredictable results. But I still can not ascertain where do the problems spring from. I mean in which, exactly, cases they do appear. So I do continue experimenting. The good thing, however, is that in many cases there are quite nice results. And this is something that it should make us content with the already reached point.

[ggaliens]

This is part of the reason by I prefer the old approach ... (sorry if offended sensibilities before ... we have all worked hard on this one ...

Code:

$ ./start_debug.sh
Eshell V7.1  (abort with ^G)
1> wings-2.0.3.1.g3601.dirty
Reading preferences from: c:/Users/markw/AppData/Roaming/Wings3D/Preferences.txt
1> Using GPU shaders.
1> wings:667: Error: {case_clause,
                   {3926,19793,
                    {edge,19793,3926,6310,6071,5260,19813,19812,19794},
                    {face_iterator,19813,6310,
                     {array,19840,100000,undefined,
                      {{{{{{edge,0,1,9934,10070,11,5,2,10},
                           {edge,0,5,10032,9871,29,2,4,30},
                           {edge,0,5272,10070,10032,10,0,1,29},
                           {edge,0,5273,10085,10647,16398,4,6,16400},
                           {edge,0,5275,9871,10085,30,1,3,16398},


%% MISSING MILLION LINES OF EDGE TABLE (Mark W.)


         {edge,19815,1086,19839,5826,5260,19815,13924,5257}},
        10,10,10,10,10,10,10},
       100,100},
      1000},
     10000,10000,10000,10000,10000,10000,10000,10000,
     10000}}}}} [{wings_vertex,connect_4,5,
                     [{file,"wings_vertex.erl"},{line,686}]},
                 {wings_vertex,force_connect,4,
                     [{file,"wings_vertex.erl"},{line,613}]},
                 {wings_vertex,connect_vs,3,
                     [{file,"wings_vertex.erl"},{line,277}]},
                 {wings_vertex_cmd,
                     '-connect_cuts/1-fun-0-',2,
                     [{file,"wings_vertex_cmd.erl"},
                      {line,464}]},
                 {lists,foldl,3,
                     [{file,"lists.erl"},{line,1262}]},
                 {wings_vertex_cmd,
                     '-connect_cuts/1-fun-1-',3,
                     [{file,"wings_vertex_cmd.erl"},
                      {line,467}]},
                 {wings_sel,fold_1,4,
                     [{file,"wings_sel.erl"},{line,84}]},
                 {wings_vertex_cmd,command,2,
                     [{file,"wings_vertex_cmd.erl"},
                      {line,64}]}]

---

And here is a link to a bunny.wings that has three vertices pre-selected and will BOMB with the 2.0.3 release ...

https://www.dropbox.com/s/kqnpbv6ngcex7a...wings?dl=0

Sorry I didn't provide something like this yesterday.

[micheus]

I found the problem you reported.
It's cause for the value assigned to the value of 0.0001 assigned to the variable Tol (tolerance) used in wings_vertex:collect_edges. It's seems to create the same issue we already discussed here in the forum related to "connect in short distance" (I simply could not found the post anywhere )

So, for your bunny the problem can be fixed just if we add another zero to the decimals: 0.00001.


I still need to check the one I reported.

[ggaliens]

A good implementation would probably be robust given either of the Tol values ... a better implementation would use a small relativistic value ... something like wings_edge:median_length(We) / 1000.0 probably a very very reasonable choice ... independent of model scale.

Just imagining what we can do to make things better.

[dgud]

Think I have fixed it now. Works on both examples posted here and many more.