v.support and d.vect plotting

Simon Cox S.Cox at solo.ned.dem.csiro.au
Thu May 2 08:00:00 EDT 1996

>Does your work with gshhstograss show that there is an improvement that could
>be incorportated into several routines used in GRASS to speed up the display
>of layers?  Has some entity taken responsibility to be a clearing house for
>new developments in GRASS?

I discussed my experience with some colleagues and we decided, without looking
inside the code, that the problem is probably in the indexing method used
by the
grass vector programs - I suspect that some kind of (exhaustive) search is
done on the polylines, repeatedly, which scales exponentially with the
number of nodes
on the polylines.  There is probably also a "bounding box" for each segment.

My solution, to break long ones up into adjoining segments, means that
using the
bounding-box for indexing is more efficient, and then searches of the nodes
on a segment
is shorter.  I have no idea what the optimum number of nodes would be and
it probably
depends on the spacing and tasks in hand.  However, it might be worth
into the "v.in.anything" programs an option to split up the long polylines
at that
stage, then the downstream problems would be gone.

To my knowledge, continued coordination of GRASS is in the hands of Rutgers
and LAS,
but neither appears to be too pro-active about this.

Simon Cox

Dr Simon Cox                        __  \
CSIRO Exploration & Mining       ,~'  L_|\    Australian
39 Fairway, PO Box 437,       ;-'         \   Geodynamics
Nedlands, WA  6009  Australia (            \  Cooperative
    T:     +61 9 284 8443    +    ___     /  Research
    F:     +61 9 389 1906     L~~'   "\__/   Centre
simon at ned.dem.csiro.au                   W

More information about the grass-dev mailing list