[GRASS5] v.in.ascii

Helena Mitasova hmitaso at unity.ncsu.edu
Wed Jun 29 23:37:49 EDT 2005


I tried to skip building topology (just commented line 416 V_build ... 
in in.c)
and the 600000 point file imported faster and did not start swapping.
However, when I tried v.info on the imported file I got an error 
message:
Cannot open old vector ..... on level 2,
so it thinks that this is an old vector file if it does not have the 
topology built?
After I ran v.build on this file (which of course does not make much 
sense for the point data)
it tried to build topology and started swapping after few hundred 
points, but finished the job.
Then I was able to run v.info.

So apparently V_build should be skipped in v.in.ascii if the input is a 
point file but then
other programs, such as v.info,  should be able to work with such file 
or there at least
should be a simplified option for v.build for points data.

Any ideas on how to improve the point data handling, fix v.in.ascii and 
V_build  ?
(if we skip topology building for points, the memory leak will still be 
there). I am slowly
learning the vector stuff but this is much more complicated that the 
old ascii sites support.
Radim, did the valgrind analysis by Hamish help or is more needed?

thanks a lot for any solution, I am sure that this will start biting 
more people than just me,
as the data sets get larger,

Helena


On Jun 23, 2005, at 8:09 AM, Radim Blazek wrote:

>> 'g.region vect=lidaratm2'  used 73mb RAM while it ran. (steady)
>
> Use Vect_open_old_head instead of Vect_open_old in main.c.
>
>> d.vect used 204mb RAM to draw (steady)
>>  (probably the next bottleneck at 8-10 million points)
>
> Try to delete 'topo'.
>
> Add -b flag to avoid building topology to v.in.ascii (in.c line 409).
>
> Radim
>
Helena Mitasova
Dept. of Marine, Earth and Atm. Sciences
1125 Jordan Hall, NCSU Box 8208,
Raleigh NC 27695
http://skagit.meas.ncsu.edu/~helena/




More information about the grass-dev mailing list