David Lehmann wrote:
The ((i++)) fails only when the result is 1.
--- What are you calling the result? The return value or the post-incremented value of 'i', which is done *after* the return-result is returned? I.e. you are talking 2 different point of return (that's one issue). Second, others have answered and that is if you have an integer 0, like if ((0)) ; then (should this execute or not?)... Since values in (()) are evaluated similarly to 'C', and in 'C', 0 usually = false... then you end up with arithmetic 0 = false = > failure case; now if you use: if [[ (0) ]]; then echo this is true; fi The wonders of shell (which make sense if you consider why, but not entirely intuitive on a casual glance...)... If you do computations in (()), be save and " ((xxx))||: ", which looks ugly after you do it a while... so you move on and forget the (()) and use: alias int=declare\ -i int i=0 int j=i++ echo $j, $i, $? 0 1 0 j=i++ echo $j, $i, $? 1 2 0 j=i++ echo $j, $i, $? 2 3 0 --- (or whatever, but the above always sets the exit code 0 so no errexit trigger)
