[GRASS-user] d.vect struggles on Map.Render() for 'larger' vectors

[Shane Litherland]

Hi GRASSers,

I've just clarified something for myself, that I've only occasionally
noticed before because most of my work is with small vector files.
Starting to play with the North Carolina dataset highlighted this issue
for me again, as it has some 'larger' files in there (not that I'd
consider 300KB coor file large!).

I've got 6.4.2svn updated 21-01-12, on Ubuntu 10.04.

The steps I've used:

1. open GRASS, in NC mapset, PERMANENT
2. click 'add vector'
3. in d.vect, choose a vector e.g. bridges
4. hit OK, and GRASS greys out for over 10 minutes, while things are
computed.

There seems to be a few pauses in terminal output,

GUI D3/5: Layer.SetCmd(): cmd='d.vect map=bridges at PERMANENT'
(the next line after about 2 minutes was
GUI D4/5: LayerTree.ReorderLayers(): items=bridges at PERMANENT,
)

but the longest (around 10 minutes) was after:
GUI D3/5: utils.GetVectorNumberOfLayers(): vector=bridges at PERMANENT -> 1

The next line after this when output resumes in the terminal is:

GUI D3/5: Map.Render() type=vector, layer=bridges at PERMANENT

Nothing showed in map display, so;
5. click 'zoom to selected layer' tool and again it greys out for
several minutes,

again the delay is at the Map.Render() step.

6. right-click vector in Layer Manager, choose 'Properties' to bring up
d.vect dialog
7. click 'Selection' tab, (it greys out again for several seconds), then
deselecting all but one Feature Type and hitting OK (starting with
points and working my way through each type)

-I tried points, then lines, but with each of them taking around ten
minutes I could not afford the time to continue through each type like
that!

I'd also tried reducing resolution - halving column/row numbers
essentially. No discernable improvement.

This issue is barely noticeable on very small files (e.g busroute1 in NC
dataset).

I've also tried a statewide 'soils' map I have in my own system, coor
file is 1.1MB. It it sluggish, but still loads much faster than the NC
bridges vector (which is a third the coor filesize). It loads
significantly faster (maybe 10-15 seconds?) if I deselect the 'area'
option on the 'Selection' tab.

So it would appear that it's not just the size of the vector file that
causes the delay?

Am I just asking too much from my computer, or is there some way to cut
down on the computing that takes place in GRASS? Should/could I tweak
certain settings for my GRASS or computer setup to improve performance?

If I can't get the NC maps to work faster than this, it's going to limit
how much I can use them to check/replicate glitches I find when doing my
own digitising etc... just this experience has chewed up about 5 hrs of
my afternoon... :-/

-shane