[GRASS-dev] [bug #3168] (grass) r.to.vect - broken pipe

Maciek Sieczka via RT grass-bugs at intevation.de
Wed Apr 25 14:52:50 EDT 2007


Markus wrote (Fri, Jul 7 2006 16:50:02):

> I think that we have this report twice:
> http://intevation.de/rt/webrt?serial_num=3877&display=History
> 
> merge?

Getting back to old stuff; good news here!

Not exactly the same. In this case there was enough RAM and SWAP to handle the
data; I didn't run out of the memory like in #3877. The problem was that
r.to.vect abnormally terminated, when it was almost done, with 'Broken pipe',
in spite of memory was still available.

As of now, I have just this re-tried the same setup with fresh 6.3 CVS and the
'Broken pipe' is gone! r.to.vect succesfully completes AND manages to create
the dbf table. Maybe there was some fix committed which I missed?

Summarising: currently, r.to.vect is able to tranform a raster created of 2 3"
SRTM tiles into vector points (2 883 600 of them!) and create a dbf table
using a little less than 1GB RAM + 1GB SWAP (merely!).

Isn't that great?

details:

$ g.region rast=srtm_odra_patched -ag
n=50.00041667
s=48.99958333
w=21.99958333
e=24.00041667
nsres=0.00083333
ewres=0.00083333
rows=1201
cols=2401
cells=2883601

$ r.to.vect input=srtm_odra_patched output=srtm_odra_patched_pkt feature=point

# Takes QUITE some time, but completes! One point less created than the number
# cells in region, cause one cell was null:

$ v.info -t srtm_odra_patched_pkt
nodes=2883600
points=2883600
lines=0
boundaries=0
centroids=0
areas=0
islands=0
faces=0
kernels=0
primitives=2883600
map3d=0

$ v.info -c srtm_odra_patched_pkt
Displaying column types/names for database connection of layer 1:
INTEGER|cat
INTEGER|value
CHARACTER|label

Closing ticket.

Maciek


-------------------------------------------- Managed by Request Tracker




More information about the grass-dev mailing list