[GRASS5] Strange r.mapcalc behaviour - bug?

[Glynn Clements]

Markus Neteler wrote:

> Oh, I see. Does this mean that the result might be ok
> but simply r.support's results wrong?

Yes. If the original map contained values up to and including 36748,
the result map will contain values up to and including 3.6748 (to
within the accuracy of double-precision FP).

> Say, If I simply
> - r.in.gdal
> - r.mapcalc
> 
> will I get correct results?

Yes.

> Then I just drop r.support from
> the above procedure. Maybe it's not needed at all, but just
> damaging the range file.

I'm not sure what purpose "r.support -r" serves. The core libgis
functions automatically update the range; AFAICT, the only reasons
that the range/fp_range files may not accurately reflect the range of
the data are:

1. The actual cell/fcell data files have been modified other than
via the standard functions (open_cell/put_row/close_cell).

2. The range has been explicitly modified via e.g. G_update_range()
(which is what "r.support -r" does).

-- 
Glynn Clements <glynn.clements at virgin.net>