BOLO: Potential shell weirdness

[Robert Elz <kre%munnari.OZ.AU@localhost>]

I have just committed fixes for /bin/sh pattern matching - the changes make the
current pattern matching ATF test succeed (all sub-tests),  as do all the other ATF
sh tests (some of which also test aspects of pattern matching, directly or
indirectly).

However, since I also wrote the pattern matching ATF test, all this means is 
that sh behaves as I think it should, for the cases I could think of to test.
It does not necessarily mean that how I think it should work is how it really
should (though I kind of hope that is not much of a problem) nor that there
are not other issues that I simply never thought to test.

So, please lsee if you can think of any other strange (or whatever) pattern
tests you can imagine (for patterns in case statements, for patterns in the
word part of a # or % (including the doubled forms) operators in variable
expansions, or good old glob (filename expansions).

Also, look out for any strange behaviour - in some ways these changes are
intrusive, and could cause other oddities.

If you see anything strange, particularly things which worked with the /bin/sh
in NetBSD current (ie: HEAD) yesterday, but no longer work today, please
let me know, or let this list know, or file a PR (or all three.)

kre