>  I've just tried to compile both CVS 1.8.1 and othersrc/cvs on my
>  1.2_BETA Sparcstation 1. Both compile without problems (though I
>  had to remove a couple of spaces on lines in macintosh/Makefile in
>  the latter), but both fail when I try to do the regression test
>  (make check). Both fail with `FAIL: branches-16'. othersrc/cvs on
>  my 1.0/i386 system happens to fail with `FAIL: 56', if that means
>  anything.
>  
>  I feel I must be missing something silly here, but I'm not sure
>  what.  Does anyone have any suggestions?

The CVS maintainers, in some form of wisdom, modified the CVS sanity 
checks so that they use features not available in most versions of the 
"expr" command.  The standard recommendation is to install the GNU 
shelltools, though a less GNU-centric observer might think fixing the 
sanity checks would be more, um, "sane".

I'm not sure if this is _your_ problem, but it is certainly _a_ problem 
with recent versions of CVS.

 - irving -