[GRASS-user] Re: r.surf.nnbathy crashed [was: v.surf.rst limitations?]

[Maciej Sieczka]

Carlos "Guâno" Grohmann wrote:
> Well, today it decided to work. I used another dataset, with ~16.5
> mill pts, with a resolution of 2.5m, ~30 mill cells. It took ~8 hours
> to interpolate.
> 
> really nice!

That's great. As to timing it performed better than I have expected. Cool.

But, first of all, I would really appreciate if you could provide the
details I asked you about in my previous email, and, if possible, the
data it failed with. I want to make my script really good (or help
Pavel Sakov fixing nnbathy), soe it doesn't fail with some data and
works with other.

A thought: maybe you just got your disk full during the first run, when
it failed? Note nnbathy creates 3 *huge* temp files - plain ASCII x,y,z
lists of the input and output, and the output converted into GRASS
ASCII format. Then it makes a GRASS raster from the latter - and only
then it removes the 3 temporary ones, when the script terminates. So at
the runtime, you need several times more space than the resulting GRASS
raster actually occupies. It is easy to run out disk space
interpolating into a grid of several mln points then, using mlns of
points for input...

Maciek