We are still testing in the IDV, but like Curtis, I was wondering if
Gridded3DSet and Gridded2/3DDoubleSet need to be updated as well.
Curtis Rueden wrote:
Hi Bill,
This change makes a lot of sense. It does not seem to have broken any of
my applications. My only comment is that we may want to consider making
the same change to Gridded3DSet, because it also does not strictly
enforce non-zero coordinate signatures.
-Curtis
Bill Hibbard wrote:
Hi Cicero,
Nasdty bug here. Your application is constructing a
degenerate Gridded2DSet and the consistency test in
the Gridded2DSet constructor is not catching it.
Your createDomain() computes the same dx (and dy)
value for i=HALFDIM and i=HALDIM+1 (j=...). Thus your
grid has grid boxes with area = 0.
I've put a Gridded2DSet.java whose consistency test
catches this at:
ftp://ftp.ssec.wisc.edu/pub/visad/untested/Gridded2DSet.java
Note that using a stricter test may break some applications
that have been getting away with constructing degenerate
grids. It is also possible that my new stricter test is
not formulated correctly - this stuff is quite complex.
So I hope the other VisAD developers may take a look at this
and test it (the changes are marked by 'CICERO' comments
in Gridded2DSet.java).
Good luck,
Bill
--
*************************************************************
Don Murray UCAR Unidata Program
dmurray@xxxxxxxxxxxxxxxx P.O. Box 3000
(303) 497-8628 Boulder, CO 80307
http://www.unidata.ucar.edu/staff/donm
"Time makes everyone interesting, even YOU!"
*************************************************************
