Re: Development version on GNU Chess

[Lukas Geyer]

Simon Waters <address@hidden> writes:

> Have removed the nominal support for UNIVERSAL board (this
> hadn't been actively maintained for a few minor releases, and
> was starting to get in the way).

Very good.

> Removal of UNIVERSAL means I could lose the test for the alarm
> system call (although this is widely available - Posix, SVR4,
> BSD 4.3 etc etc) in configure, remember to add it back in if you
> ever need it.

At the moment, these tests are very dubious anyway, because we don't
do anything in case of failure. And whether the test is there or not,
it will fail while compiling if one of those functions is missing. We
should probably see if we can work around things in the code, or just
fail in the configure script. I would prefer the latter, and most of
these functions are so standard that nobody would notice anyway.

> Also since test release of 5.05 I have cleaned up the format
> modifiers for printf output, so they are now not only working
> correctly, but logically correct, and should compile on stricter
> compilers than GCC.

Yes, just tested with Sun's cc, compiles flawlessly, no warnings or
such.

> Since the test release we have new EPD parser and correction for
> buffer overflow in pgn.c
> 
> I want to get 5.05 out tomorrow - I'll probably tidy up the
> NEWS, and similar text files, but I'll only accept security
> related changes, or changes required to compile on previously
> supported platforms that have been broken by other changes, or
> complete showstopper bugs, for 5.05.

I did not find any serious bugs, but there is

DISTCLEANFILES = *~ GCint.h

in Makefile.am, whereas it (well, the GCint.h) belongs in
src/Makefile.am. (Alternatively, replace the above by src/GCint.h.) If
you can change that before the release, it would be fine.

>  This one needs to get out, so we can move on to more exciting
> changes.

Agreed, we should really start to get the AI engine going, so you can
talk to gnuchess during games and distract the program...

Cheers,
Lukas

-- 
This is not a signature