[GRASS-user] problem with r.li. modules-Charakter Encoding?

[Glynn Clements]

Martin Besta wrote:

> I have noted one little problem with r.li. modules, when trying to use
> overlayed vector map as sampling areas. As example, r.li.edgedensity writes
> back this message:
> 
> r.li.patchnum map=r1953_10m at landuse conf=pokus1953 output=pokus1953patchnum 
> 
> Illegal filename. Character <ï> not allowed.
> Illegal filename. Character <ï> not allowed.
> Illegal filename. Character <ï> not allowed.
> 
> Can't open header file for [(Qï·PQï· ?¿ in (null)]
> 
> CHILD[pid = 11280]: unable to open (Qï·PQï· ?¿ mask ... continue without!!!

> I guess, this can be caused with incorretct character encoding (my current
> settings is ISO-8859-2),

I doubt it. It looks as if an error is occurring, and rather than
detecting this and aborting, r.li continues processing but using
garbage data.

Unfortunately, r.li is quite complex, and the fact that it is
implemented as several distributed processes can make debugging
awkward.

> The second problem with r.li modules is, that I cannot calculate any indice
> on 1 m resolution raster with any r.li module. The same, 10 m resolution
> raster goes great...is 1 m resolution supported by r.li modules?

r.li is unlikely to care about physical resolution (in metres), but it
may well fail if the total number of cells is too large. This will
impose a limit on the combination of the region size and the
resolution, i.e. you may be able to use 1m resolution on a small
region, but not on a larger region.

-- 
Glynn Clements <glynn at gclements.plus.com>