AWK subtract numeric record two from record one

Question

I cannot solve a very simple problem. My data file looks like:

Crap Crap 0.123456789D+09 Crap Crap
Crap Crap 0.123456798D+09 Crap Crap

I need to use AWK to subtract the number in the third column; the SECOND row minus the FIRST row.

I tried:

cat crap.txt | awk '{ A[NR-1] = $3 } END { print A[1] - A[0] }'

to no success. Maybe the format of the number is wrong? (Can AWK read scientific notation with D instead of E?)

Help!

EDIT:

Just so the community knows, AWK does not understand scientific notation which uses D instead of E (like many Fortran outputs produce). It is necessary to replace the D for E and then carry on any mathematical operation.

Answer 1

You could change the notation on the fly...

cat crap.txt | awk '{ sub(/D/,"E",$3); A[NR-1] = $3; } END { print A[1] - A[0] }'

Answer 2

awk cannot read D-notation (what does it mean by the way?). Here is an example:

printf "1.2D+1\n1.2E+1\n12\n" | awk '{ print $1/3; }'
0.4
4
4

The first one is wrong.