Re: pkg/32263: nvi-m17n checksum mismatch

[Curt Sampson <cjs%cynic.net@localhost>]

The best thing is, some sites have the one binary, some the other, so
depending on your pkgsrc configuration (MASTER_SITE_OVERRIDE, etc.) and
connectivity, you may still get one or the other. In Japan, I seem to
pretty consistently get the old one, not the new one (in terms of when
the checksums were committed).

I wonder, do we want a way to deal with multiple checksums? Or do we
perhaps want to try to chase down all the holders of one and get them to
switch to the other? Or just rename the source tarball, put it on the
NetBSD servers, and make everyone get it from there?

At the very least, we could chase down the Japanese copies and get them
to match.