[GRASS-user] v.patch fix/workaround needed
peter.loewe at gmx.de
peter.loewe at gmx.de
Wed Aug 26 10:20:24 EDT 2009
Hi
> >
> > In a first step, a polygon ("A") is created using v.patch from an
> outline and a centroid. This works fine.
v.patch: "A-outline" + "A-centroid" => "A-polygon" (-a Option)
> >
> > In a second step, the newly created polygon is added to a larger set of
> polygons ("SET") using v.patch again.
> >
v.patch -a "A-polygon" + "Set" => "Set"
> > In a few cases (reproducable) the "SET" polygon is NOT extended by the
> proper polygon "A" but the centroid of "A" and the polyline of "A" (the
> process works correctly in 9 out of 10 cases).
> >
> > Since this process is to run as a script, how can the disjoint tuple of
> controid-A and polyline-A in the "SET" vector layer be mended again ?
>
> Before we start hunting bugs for nothing, could you please try
> GRASS 6.4? AFAIR there have been implemented some fixes in the
> topology engine... perhaps it already works?
>
> Markus
The results are the same on Grass6.4RC5.
I was able to track down to different phenomena.
Some background info:
The dataset "Set" consists of many squares (all imported as as "A-polygons"). SOME OF THEM OVERLAP.
1)
As long as the centroids of the affected squares are OUTSIDE of the overlapping area, everything is OK.
If the would centroids fall INSIDE the overlapping zone, the newly to be added polygon is only vpatched as a line and a controid.
Since the intentional use of overlapping polygons is surely a bad idea, this makes sense. However, some sort of feedback to the user would be a good thing to have if such a case occurs.
2)
Yet, there are cases without "critical overlap" resulting in centroid/line-pairs: In each case two identical centroids (holding identical attributes but different categories) were created.
-->This could point to a bug in v.patch.
Peter
--
Dr. Peter Löwe
<peter.loewe at gmx.de>
Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3 -
sicherer, schneller und einfacher! http://portal.gmx.net/de/go/atbrowser
More information about the grass-user
mailing list