I’m trying to stop points being counted twice when using a series of calls to gmtselect with a solid grid of squares. Points that are exactly on the border can get selected in adjacent grid cells.
So I want to add a tiny fraction to each lat and long (0.00005) to stop this (the lat and longs are only specified to 4 decimal places).
My command is of the form:
gmtselect -Fpoly.txt -i5+o0.00005,6+o0.00005,0-7 -o2-9
The required adjustment is being made to column 5, but not to column 6 - it just gets passed as is.
What am I doing wrong?
Using gmt 5.3.4
0-7 is resetting what you have for 6
Then why does 5 get the adjustment?
Not sure but you are specifying both 5 and 6 twice
The manual states that columns may be repeated. I would have thought the repetitions of a column were “independent” and hence my original construction.
I did a clunky work around using:
-i5+o0.00005,6+o0.00005,0,1,2,3,4,7 -o2-6,0,1,7
but this meant that the lats/longs that were passed were the scaled ones, not the original. No big hassle, but annoying.
Should the column independence be a feature?
No, repeats are allowed. I see, you intentionally wanted what you wrote first. I will see if I can recreate the issue tomorrow.
If I do this:
echo 0 1 2 3 4 5 6 7 > t.txt
gmt select t.txt -i5+o0.00005,6+o0.00005,0-7
I get
5.00005 6 0 1 2 3 4 5.00005 6 7
So you are right there is a bug in that column 6 is not affected by the offset; I will look at that. However, because each input column can only have one scaling and offset, it is not possible for you to write out the same column with two separate offsets.
This bug has now been fixed in the repository and the output of my test now gives what you expect:
gmt select -i5+o0.00005,6+o0.00005,0-7 t.txt
5.00005 6.00005 0 1 2 3 4 5 6 7
Thanks for pointing out the bug!
Excellent. Thanks Paul.
Love GMT!