Once again it's reached the point where (at least on SPARC Solaris 8
with all but a couple of the latest patches installed) nawk fails
during patchadd with too many fields.

My usual workaround is to temporarily move nawk out of the way and put in
its place a symlink to (a self-compiled copy of) GNU gawk, whose design
has less in the way of arbitrary limits (given sufficient memory, I
suppose).

How about making that workaround into a real solution - include gawk
in the same package as nawk (or one even more likely to always be
present), and use gawk in the patchadd script. Alternatively, for
Solaris 8 and later, where /usr/perl5 is presumably always present,
why not rewrite the entire patchadd into a perl script, which would
almost certainly be faster. Either of these would avoid the need for
patch nawk yet again to increase its limits.

--
mailto:rlhamil@mindwarp.smart.net [url]http://www.smart.net/~rlhamil[/url]