Jan

On the advice of Angus I did not (yet) put any effort in to group spatially.My data are already organised (spatially) fairly well.So, I'll give a few details first which have to do with my library/organisation.I use a format for a a collection of polygon with holes. In my lingo that is a MultiComplexGon. The data comes as shapefiles, which are read into a single object as before, and then gets sent to the dissolve routine, which puts the data into Path objects, calculates and puts the result back.dissolveloadfactor...

On the advice of Angus I did not (yet) put any effort in to group spatially.My data are already organised (spatially) fairly well.So, I'll give a few details first which have to do with my library/organisation.I use a format for a a collection of polygon with holes. In my lingo that is a MultiComplexGon. The data comes as shapefiles, which are read into a single object as before, and then gets sent to the dissolve routine, which puts the data into Path objects, calculates and puts the result back.dissolveloadfactor...

On the advice of Angus I did not (yet) put any effort in to group spatially.My data are already organised (spatially) fairly well.So, I'll give a few details first which have to do with my library/organisation.I use a format for a a collection of polygon with holes. In my lingo that is a MultiComplexGon. The data comes as shapefiles, which are read into a single object as before, and then gets sent to the dissolve routine, which puts the data into Path objects, calculates and puts the result back.dissolveloadfactor...

On the advice of Angus I did not (yet) put any effort in to group spatially.My data are already organised (spatially) fairly well.So, I'll give a few details first which have to do with my library/organisation.I use a format for a a collection of polygon with holes. In my lingo that is a MultiComplexGon. The data comes as shapefiles, which are read into a single object as before, and then gets sent to the dissolve routine, which puts the data into Path objects, calculates and puts the result back.dissolveloadfactor...

Hi Angus, I tried smaller groups, but they gave significantly worse results. 1000 was a good start, but also to my surprise, higher was better but 5000 too much. My multithreading does not cover this part, it would have to be nested multithreading, I'm looking into that, as it would help with the largest datasets. Also I haven't spent energy to find out whether the clipping of the groups individually or the clipping of the groupresults takes more time. I hope I can select a smallish sample, but first...

I've implemented subdivision and tweaked it a bit. In my task the effect is substantial, from about 8 hours to 40 minutes (factor 12). Subdivision is by a factor of the number of items (per class), which varies quite a bit. That worked better than using fixed numbers of polys per subtask, say 1000. I have not preordered the polygons, but they have some ordening stemming from a set of shapefiles covering regular tiles of the whole area. That may also explain why a fractional size worked better than...

Hi Angus, Was caught up in some socialities ;) I used the PolyTree approach years ago because it gave me the holes. If you say I shouldn't need to do this, I'll have another look. Now I'm converting to 2.0 and have a few loose ends: ReverseSolution(), SimpliFyPolygon(), CleanPolygon(), StrictlySimple() are no longer there, I'm trying to find out whether I really need them, but I'm not comfortable to scrap them; less important but potentially so: ArcTolerance is no longer available and tuning variables...

Thanks! On Sat, Nov 2, 2019 at 11:03 PM Angus Johnson angusj@users.sourceforge.net wrote: Do you have any reason to believe that the preview version is less correct than the preceding version 4.6.2? (I'm not asking for guarantees, mind you) I'm confident it's still just as robust as version one. However, it's still only beta because there are additional features I hope to address such as removing micro self-intersections related to rounding (which also occur on version 1) and a more complete solution...