Re: GNU Chess

[Simon Waters]

"David A. Wheeler" wrote:
> 
> Hi - I'm thinking about trying to make
> some modifications to GNU Chess, but I don't know
> if you guys are actively accepting improvements, or if
> you'd prefer I worked through CVS, or whatever.
> 
> PLEASE let me know if you are actively considering improvements to
> GNU Chess 5.04; I don't want to waste my time.

We are working from the CVS head on Savannah, which has some
boring global changes of short to int, and similar, so better to
work from CVS head.

http://savannah.gnu.org/projects/chess/

Contributions welcome but at the moment for version 5, I have
commit privileges, I usually get patches as a diff from CVS
head, check them for sanity, and commit them. I'm not CVS, or C
guru, but it should make it easy for others to stay upto date.

If you let me know before you cut code, or make major changes, I
can tell you if anyone else has told me they are working on it.

The only big thing to know about currently is Lukas was working
on pondering, which is likely to rework the whole of cmd.c, and
the transposition tables.

You should be able to do the PGN code seperately I think.

If I become a major bottleneck we can give commit privileges to
others, although I have tidied up the last few patches, and try
to keep the documentation in sync.

> In particular, the current PGN reader is awful.
> I even found a case where it can't read back in a PGN file it saves,
> and it doesn't accept a lot of PGN syntax
> (% commands, ; comments, {} comments, RAV, and so on).

> It can't even handle PGN where the move numbers are omitted, or where
> FEN is included, or where black starts.
> It also can't handle "nonstandard" but common PGN files
> (Mac or Windows line endings, KxP format, etc.).

I think the only thing I'm unclear on is black starts, I know I
fixed a black starts bug somewhere, maybe it was on output from
a EPD position.....
 
> I'm thinking of rewriting the PGN reader (both book and non-book).
> I might use flex instead of continuing the "hand scanner" approach.

flex is fine, I don't particularly have a problem with requiring
flex in the build, but if we can ship enough prebuilt code to
avoid requiring it that is better still.
 
> BTW, here's the file that it saved and can't reload; trying to
> pgnload it produces the error:
> Illegal move 6. ... Bb4+

5.05 also gives the same error, although the move is loaded.

I was planning on releasing the CVS head after tidying up some
format modifiers, but in the light of the buffer overflow
problems it might be wise to fix the parsing of PGN, FEN and
commands as well. Although some of this code will go when Lukas
perfects his pondering patch.

 Simon