Hi,

In terms of a solution for picking one (and only one) version of awk to use, the problem here is you're doing three checks that are un-related to each other, and (as you've found) if multiple awks are found then you can end up with undesirable behaviour or the 'wrong' awk being picked.

The best thing to do here I think would be to use an if/elif/else kind of decision process, so that based on your order of precedence you pick one (and only one) version of Awk at all times.

For instance, if your order of preferred awk implementations is:

then the following code would ensure that one, and only one, was picked (and it would be the first one found, in order of precedence):

Code:

if [ -x /usr/bin/gawk ]
then
        AWK=/usr/bin/gawk
elif [ -x /usr/bin/nawk ]
then
        AWK=/usr/bin/nawk
else
        AWK=/usr/bin/awk
fi

So even if your system has both gawk and nawk, it will always pick gawk and leave it at that.

Hope this helps.