Re: bad error handling in pairs-imp.c

    > From: Matthew Dempsky <[email protected]>

    > In the printing routine for pairs, there were a few bad goto
    > statements.  patch-18 in my branch fixed those (I'd write a unit test
    > too, but I'm not sure how to cause printfmt to fail intentionally).


Dude!  

I'm now caught up with merging in your changes, none of which required 
any corrections, and which in general displayed that you are pretty 
damn closely reading the source and catching things not caught by the 
unit tests.

So, thanks.

Don't obsess about extending the unit tests for bugs that you find.
They aren't regression tests.  There's no testing religion here.
They're just a quick-and-useful sanity check and when it's convenient
to extend them, great -- when not, oh well.

-t