On Thu, Jan 05, 2006 at 07:52:02PM +0100, Havard Eidnes wrote:
> 
> Well, we don't want to introduce a dependency on perl in our
> build system, now, do we? ;) However, the first part is simple
> enough that it can be done in awk, and the second part can be
> done by piping the result into sort, and, yes, I think it would
> be a good idea to do something along those lines.

You can't use sort, because you need the last copy of duplicate lines.
I have a shell script that collapses METALOG, but it won't run with
many shells - so will be a problem for non-netbsd builds.

Also a METALOG file from update builds can easily contain information
for files that no longer exist.  This has not been a problem in the
past (except when a file changes to/from a symlink), and it needs to
remain so.

	David

-- 
David Laight: david@l8s.co.uk