Hi everybody,
yesterday Markus has commited a bug fix and
enhancement version of r.contour that I have hacked during the weekend.
What's new:
* category support bug has been fixed: r.contour2 produces also
dig_cats file, so d.what.vect no more complain about missing
category support. Since a contour level could be floatig point
I've inserted in the label the true value, whilst category number
remains its integer equivalent ( ===> so there may be problems if
one asks for fp contour levels that match the same integer number, I
haven't tested that case);
* a little trick (derived from GMT contour program) avoids some
kind of backtrack (well, I was thinking any kind of backtrack, but
Justin showed me that I was wrong : data matrix is parsed,
and if a cell whose data matches exactly
a contour value its value is added a small quantity (the smaller
one that makes cell[i] + small != cell[i], I've taken into
consideration double encoding).
* one can specify a minimum number of point for a contour line to
be put into outout, so that small spurs, single points and so
on won't be present and the output will be more clear (that's
optional anyway);
* in my opinion there were minor memory handling problems in
r.contour, I've corrected them (Head.map_name was not guaranteed
to be properly terminated, Points structures in contour function
were not deallocated).
Markus has corrected some remaining bugs as well.
Bye
Andrea Aime
----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'
I grabbed the r.contour yesterday and gave it a test run. It seems to
work quite nicely.
I just have a comment or question about compiling.
Of course it compiled fine under Linux. It gave me a bit of a problem
under Solaris (7). The problem was the definition of DBL_EPSILON. I
believe this is defined in float.h. The float.h file is referenced under
Linux from math.h which is good. It is not referenced by math.h under my
Solaris. I simply added include float.h to the source (main.c) and it
built and worked fine on Solaris.
Could the definition of DBL_EPSILON be added directly to the code to
avoid dependancies or does its definition vary between platforms?
--
Bob Covill
Tekmap Consulting
P.O. Box 2016 Fall River, N.S.
B2T 1K6
Canada
----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'
Justin Hickey already did that.
Just grab again r.contour from CVS, but beware, we are still
working on it, so in the next days we may release a new version.
(btw, if anyone wants to collaborate he would be welcomed )
Bye
Andrea Aime
Bob Covill wrote:
Andrea,
I grabbed the r.contour yesterday and gave it a test run. It seems to
work quite nicely.
I just have a comment or question about compiling.
Of course it compiled fine under Linux. It gave me a bit of a problem
under Solaris (7). The problem was the definition of DBL_EPSILON. I
believe this is defined in float.h. The float.h file is referenced under
Linux from math.h which is good. It is not referenced by math.h under my
Solaris. I simply added include float.h to the source (main.c) and it
built and worked fine on Solaris.
Could the definition of DBL_EPSILON be added directly to the code to
avoid dependancies or does its definition vary between platforms?
--
Bob Covill
Tekmap Consulting
P.O. Box 2016 Fall River, N.S.
B2T 1K6
Canada
----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'
----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'
: data matrix is parsed,
)