[GRASS-user] Re: feedback on v.vect.stats

[Markus Metz]

On Fri, Dec 10, 2010 at 5:36 PM, Patrick_schirmer <patrick_GIS at gmx.ch> wrote:
> Hi Markus,
>
> A short Feedback on the add-on "v.vect.stats" that you created.

Thanks!
>
> Had an error saying "segmentation fault" for various reasons that were not
> precised any further:
> Thought you might want to have a feedback eventually. I can't read
> "C"-lanuage and can't check the code. If this is the wrong place to give a
> feedback like this, please tell me where I should post this in future.
>
The mailing list is the right place.
>
> 1.) columns "pcolumn, scolumn" are not created if not existing =>error is
> "segmentation fault"

I guess that is partially a typo on your side because pcolumn is never
created, it must exist because it is the column in the point vector
from which stats are collected. The segfault when creating ccolumn
and/or scolumn should be fixed as of r44583.

> 2.) my "scolumn" was called "d_MEAN" in Postgresql. It seems postgresql does
> not recognize UPPER CASE. When I mentioned scolumn in the command with a
> upper case, error would be "typ must be double" instead of mentioning that
> the column was not existing

I tested only with dbf and sqlite, it's now working there. Could you
please test the fixed version with Postgresql backend? Thanks!

> 3.) pcolumn must be integer (not numeric). It will create error
> "segmentation fault" if not.

Should be fixed by now (r44583).
>
> Then it worked great for points. (If you have a hint what approach I could
> take to perform the same analyses for lines that are completely in a polygon
> it would be great by the way).
>
I was thinking about that too, extending it to collect stats on
points, lines or areas. That would be a bit of work, you could create
am enhancement ticket in trac as a reminder...

Markus M