It should be there because nawk is not enough, not always.

There are some gawk extensions that NetBSD's nawk does not include,
or where nawk is not quite compatible.  gensub() is one (if you
pass in a numeric parameter, nawk's gensub() is off-by-one relative
to gawk).  Another is that I believe nawk has a limited number of
concurrently open files, while gawk does not (or has a much higher
limit).

Another, harder-to-fix incompatibility is in gawk's gensub() where
you can refer to the n'th parenthesized expression from the regular
expression part.  This simply is not (last I saw) supported by nawk.


I ran into all three of those in a single script shortly after NetBSD
dropped gawk and switched to nawk.  I would be surprised if there
weren't more issues.  Whether or not knowledge of those issues is
the reason for requiring gawk, I don't know, but those issues are
reason enough to say that nawk is not always enough.



Just my $0.02718281828459045... cents' worth.


-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/