Hi,
In discussion below
- "pg" = /usr/games/polyglot version 2.0.4
- "usr" = /usr/games/xboard version 4.9.1
- "home" = /xboard-6793017/xboard version 4.9.1
On 10/27/25 03:56, h.g.muller wrote:
You get a tar ball when you do not ask for the 'tree', but for the
'snapshot'.
When you do have selected the tree, on the GitWeb interface, you could
also clock 'raw' for book.c, which will download or display that file
(which is the only file I changed).
Beware that you should be in branch v4.9.x; this is where I pushed the
chances too.
Okay I managed to build this, it was quite a saga. But maybe I did not
build it correctly. See if I missed anything.
* Install dependencies texinfo and librsvg. Then:
$ tar -xzf ~/Downloads/xboard-6793017.tar.gz
$ cd ~/xboard-6793017
$ ./autogen.sh
## I couldn't get the PKG_CONFIG_PATH to work, so I did it manually.
$ ./configure CFLAGS="-I/usr/include/librsvg-2.0"
LIBS="-L/usr/lib/x86_64-linux-gnu -lrsvg-2 -lm"
CAIRO_CFLAGS="-I/usr/include/cairo"
CAIRO_LIBS="-L/usr/lib/x86_64-linux-gnu -lcairo"
$ make xboard
There were a _lot_ of warnings. I looked at them and saw deprecated
features, misleading indentation, and array sizes. I did not see any
warnings that indicated something missing from the build.
I had to update ~/.xboardrc
;-pieceImageDirectory="~/xboard-6793017/svg"
-pieceImageDirectory="/usr/share/games/xboard/themes/default"
$ ~/xboard-6793017/xboard -debug -ncp -settingsFile /etc/xboard/xboard.conf
Good news 1. In the Reh1 position the Edit book window now shows what is
in the book.bin data. But you already knew that.
25.0% 1 Rf1 {1/1}
25.0% 1 Rg1 {1/1}
25.0% 1 Rg1 {1/1}
25.0% 1 Rh1
Good news 2. Save Games as Book gives the same results for "usr" and "home".
Bad news 1. Doing diff on xboard.debug(s), you have lost the messages
"(2-type) castling 4 2" for both 9.O-O-O and 15...O-O-O.
Bad news 2. Sometimes when moving forward in the book, the Edit book
window is empty. And each time that happens in the GUI, in the terminal
I get a message
Gtk-CRITICAL **: 03:58:39.183: gtk_text_buffer_emit_insert: assertion
'g_utf8_validate (text, len, NULL)' failed
Moving backward in the book it doesn't happen, except for 2.exd5. I
checked "usr" and it also shows an empty window for 2.exd5. I'm not sure
why I didn't notice that before.
Bad news 3. Sometimes when moving forward in the book, the Edit book
window has a long-algebraic move like "e2e4=M". The character after the
= sign seems to be a random UTF character. Right-clicking on one of
these moves gives the same Gtk error. Again, moving backwards in the
book it doesn't happen.
Hexdump forensics.
I removed the Re1-h1 game and did "pg" make-book and "usr" Save Games to
Book. For the "usr" hexdump I modified (weight,learn) to "0001 00000000"
so I could compare (zobrist,move).
polyglot is storing chess960 castling move (ah) in the book. Yuck.
xboard is storing normal castling move (cg).
In limited testing, scid is expecting normal castling move.
As you noted, xboard is dropping the last positions (4 in this test, 6
in the previous email).
adbennet61@ADB23LENZ13MX23:~/Documents/tmp
$ diff book-2-pg.txt book-2-usr.txt
## i reformatted the diff
< 048aacfc20ea2174 0f38 0001 00000000 ## e8a8
> 048aacfc20ea2174 0f3a 0001 00000000 ## e8c8
------------------------------------------------
< 337bfb7b19830fba 0ffc 0001 00000000 ## h8e8
< 3b399512a6590540 0184 0001 00000000 ## g1e1
------------------------------------------------
< 969734aebd796a89 0100 0001 00000000 ## e1a1
> 969734aebd796a89 0102 0001 00000000 ## e1c1
------------------------------------------------
< e5994713bd0c3d66 0ffc 0001 00000000 ## h8e8
< eddb297a02d6379c 0144 0001 00000000 ## f1e1
> ffffffffffffffff 0000 0000 00000000
To test for an internal limit, I took a 77-move game (the same game
twice so there would be a game list), did "usr" Save Games to Book, and
got 154 rows/ply (plus the trailing ffffffff record).
--
Alan
