[GRASSLIST:5700] Re: 6.0beta2: Input sites?

[Moritz Lennert]

On Wed, February 9, 2005 18:05, Jose Luis Gomez Dans said:
> Some extra information:
> I think this problem might be memory related. The
> following input works
> as expected:
> head -n $N XXXX.xyz | v.in.ascii output=xxxx_s x=1 y=2
> z=3 columns='x
> double, y double, z double, tol double' cat=0 -z
> If I set $N to 200000, it works. The memory usage
> jumps to around 75% of
> my memory (512Mb), but it works, which in other words
> means that v.info
> map=XXXX -c returns:
> INTEGER|cat
> DOUBLE PRECISION|x
> DOUBLE PRECISION|y
> DOUBLE PRECISION|z
> DOUBLE PRECISION|tol
>
> as expected. However, If I want to import my
> collection of points, disk
> swapping occurs (Debian Linux, btw), and it takes
> forever. Then, my
> problems (as described in my previous message) arise.
>
> Looking at the output from top, it appears that the
> memory hungry
> program is dbf, which behaves for smaller number of
> points.
>
> A temporary work-around (assuming that it is indeed
> memory that's
> corrupting something) would be to load the files by
> bits, and then
> paste them together, but I don't know how to do this
> (no v.* command
> seems appropriate).
>
> In 5.7 (mandrake), I can load the file without dbf
> taking over my whole
> memory.
>
> Any ideas?

This bug has already been reported:

http://intevation.de/rt/webrt?serial_num=2903&display=History

Moritz