🚀 go-pugleaf

Here's some queries about gnuchess 4.0 PL 69:

1. I thought the book and test games were going to be packaged separately from
   the gnuchess source code. Although there are separate .tar.gz's for the
   book and test games on prep.ai.mit.edu, the gnuchess .tar.gz file still has
   both the book and test games in it !

2. The gnuchess.1 man page in the doc directory needs updating. It claims that
   "gnuchess -c 22" gives you a 4MB hashfile, which is wrong - it gives you an
   order of magnitude bigger hashfile ! "gnuchess -c 17" (on our HP 9000 Series
   700 systems) gives you a 4MB hashfile. Also the man page nroff source starts
   with: .TH gnuchess 6 (implying a Chapter 6 man page), but the file is
   called "gnuchess.1" !

3. Am I the only one fed up with those CL* files getting created in the current
   directory ? I finally scoured the source code and changed line 702 of
   dspcom.c to read:

   sprintf (fname, "/tmp/CLp%s.%s%s-%s%s%s%c", patchlevel...

   This saves CL* files in /tmp, which is where they SHOULD go !

4. I'm still getting literally hundreds of ANSI C warnings about signed char *'s
   being used as params to routines like strcpy() etc. Considering no UNIX
   system I know of prototypes strcpy() using signed char * (they use just
   char *), surely this is going to hit a lot of users (except SunOS, who are
   so far behind everyone else w.r.t. ANSI C prototyping in system header files,
   it's laughable) ?

5. This prototyping clash comes to a head in uxdsp.c (Lines 49-51) and
   nuxdsp.c (Lines 50-52), which prototypes strcat(), strcmp() and strcpy()
   to use signed char *'s, which is a clear UNIX error. Either the prototypes
   should be removed or changed to char *'s.

Richard K. Lloyd, 
Computer Science Dept.,   JANET     : rkl@uk.ac.liverpool.compsci
Liverpool University,     Internet  : rkl@csc.liv.ac.uk
Merseyside, England,
Great Britain.