[GRASS5] Re: [bug #3826] (grass) two r.resamp.rst issues

this bug's URL: http://intevation.de/rt/webrt?serial_num=3826
---------------------------------------------------------------------
Subject: two r.resamp.rst issues

If you run r.resamp.rst at a resolution other that that of the input
map, the output elev map is not created at the specified ew_res= and
ns_res= resolution.
A warning is given that you are doing so, but it doesn't mention why
that might be bad. Interestingly the output map is actually written at
the target resolution, but it contains the data resolution from the
original map. (target was 50m, original was 100m)

If the input map is in another mapset (map@othermapset) then it gives
an error that the input map can't be found. (happens when it tries to
copy the color table I think) and exits with an error. (but processing
is already done so output map is ok, just without a colortable)

Hamish

ps - I've tidied up the module's code and strings a bit in CVS, so cvs
update before fixing...

Michael wrote:

We've run into that here too. One of my RA's did a test of
interpolating to a higher resolution using r.resamp.rst and the more
complicated procedure of generating a large number of points from a
raster map, and then interpolating from the points to a higher
resolution.

r.resamp.rst is faster and more straightforward. BUT (a VERY big but),
it behaves exactly as you describe. It simply turns out a map that
looks just like the original map is subdivided into more grid cells,
without any interpolation. This makes the module essentially useless.

Jaro wrote:

It's not useless because if you run the module at a resolution equal
to the input raster map, the result is OK.
The warning message should be replaced by a fatal error that exits the
command.

Ok, done in 6.1-cvs, but would a better long term fix be to address why
the region switching code alreadyt in the module doesn't take care of
this for you automatically? Maybe the first thing the module needs to do
is switch to the input map's resolution (internally). Perhaps that is
not in line with the way most other GRASS raster modules work so should
be avoided?

The colortable error in IL_resample_output_2d() is still there.
(G_open...() needs to be one of the other G__open..() fns so the mapset
isn't stripped off at the "map exists" check, or the mapset that is
returned from the G_open..() function needs to be passed to IL_resam..()
?? I may have my fn names wrong, but the fix should be along those
lines.)

Hamish