vancleef@ohstpy.mps.ohio-state.edu writes:
> Yeah. Part of the problem is that over the past years FIBS has gotten
> ALOT of support from developers and users, the end result being a
> state-of-the-art BG server with tons of protocols. Of course, I
> don't know what those protocols are! I have an old 'raw board'
> protocol, but I don't know if it has been changed. I will be happy
> to incorporate any protocols, as long as someone tells me what they
> are.
>
> Secondly, like I said before, I intent on donating the NetGammon code
> to the Net (with the normal "if you use my code to make money I am
> entitled to royalties" restriction) in the near future, I so believe
> it or not, it is worth spending your time helping me to develop
> to whatever 'standards' are out there these days.
>
> -Garrett
Garrett,
One thing that FIBS doesn't have that you (or fibs) could implement that
would be nice is a way of obtaining a copy of a MATCH in STANDARD format.
The way that GEnie implements this is nice: They archive all completed
matches for a period of about 1 week. Anybody may download a match
of interest during that time frame. Besides ones own matches, it would
be great to download matches of very good players after they had been
played for personal study, or to obtain large collections of such matches
for statistical study.
If you have any modern backgammon match books about, you can get an
idea of how matches are represented in print. For computer matches,
numbering each move, with one move per line is probably easiest to
impliment. Briefly:
-- The board is numbered from 24 to 1 from the rollers point of view.
ie: the numbering changes every turn, with the lower numbers
represing the rollers home board.
-- If a two column format is used, the players are in the same column
in sucessive games, regardless of who moves first. If they are assigned
'X' or 'O' or 'White' or 'Black', they maintain this designation for the
entire match
-- Dice rolls are a pair of numbers, often with followed by a ':',
sometimes in parenthes. ie: 31: is a roll of 3 & 1. I personally
prefer NOT to seperate rolls (or moves) with dashes, as when match info
is cut and pasted into word processers, sometimes the information is
broken accross lines. Non rolls (closed out) are often indicated by two
dashes ie: '--:' or '--'.
-- Moves:
Some people use 'landing point' format, while others use 'from' notation.
And opening 32: would be written as '5 5' in landing point notation, and
as '8/5 6/5' in from-to notation. Again I'm showing my preferance against
'-' as a seperator. If a checker is using more than one die in a move
the intermediate landing spot doesn't need to be represented. ie:
'64: 24/14' (from-to) or '64: 14' (landing pt). It certainly doesn't
hurt to write the entire move: '64: 24/18 18/14' This may also be
written as one move: '64: 24/18/14'.
Non Moves are often written as '-' ie with 2 on the bar: '61: B/24 -'
or '61: 24 -'. '66: - - - -'.
-- Moving mulitple pieces at the same time may be represented by adding
a multiplier in parenthes after the move. ie: '66: 24/18(2) 13/7(2)'.
or '66: 18(2) 7(2)'. Again, to aid in the simplicity of the format
all four dice moves may be indicated seperately.
-- Illegal moves need to be accounted represented and indicated.
-- Hitting. Hitting should be entered as part of the move with an '*'.
ie: making the 5-point and hitting a blot there: '31: 5* 5' or
'31: 8/5* 6/5'. A pick an pass in from-to: '65: 24/18*/13'.
-- Bearing off is usually represented by making the dstination 'O'.
ie: '12: 2/O 1/O' or '12: O O'. The number '0' may also be used in
place of 'O'.
-- Doubling and taking/passing should be counted as a move, and numbered
according to however may columns are being used (1 or 2).
-- Games should have the match score before the games begins, and a way
of determining the number of points won at the end.
-- Matches should have all of the pertinent information about players
names, Location, Date, match length available.
-- Formats. I think that it is best to encapsulate all of the games
within a match and the match info within one file. I've seen one
otherwise very good system take up enourmous amounts of hard disk
space as each 100-200 byte game took up a 16K cluster! A couple hundred
matches is only about a meg, but it takes 30M of disk space. Cluster
sizes tend to rise with disk size on DOS machines.
'Landing Point' only format with full abbrieviations in two columns
is essential for recording live matches with pen and pad. Most printed
matches are in 'from-to' format, also in two columns with full
abbrieviations. For computer generated, computer readable matches,
it could be argued that the simplist form ought to be used. IMHO this
would be a single column format, with each move numbered, and each
move of the die represented indivudually in the from-to system.
This is the easiest to parse, the easiest to convert to any other
system. For printed output, I believe that the 2-column, line numbered,
fully abbriviated, from-to notation is best. This leaves out any
ambiguity that might arise with the landing-pt only format.
Perhaps it is time to form a formal standard(s) for matches which will be
archived electronically. This standard(s) could be used for the archival of
matches played on electronic servers, computer games, as well as matches
played in the past and present and hand recorded by backgammon scribes or
recorded on video.
Once a match/game archive standard has been defined, it is likely to
find its way into programs capable of saving matches (clients, servers,
games, nn, etc...). Programs capable of importing games/matches might be
created to play back, hand annotate, or evaluate matches. Imagine being
able to feed your last match into a (now unavailable) expert level neural
net program, and have it's opinion of your 10 worst plays ready the next
morning?
After a database of high level matches becomes available it becomes
possable to do statistical searches of the database. Information like
the best possable 41: in 20,000 actual games, or the best responces
to opening rolls from a historical perspective would become available.
Soo...
Is a 'formal' standard electronic match archivel format(s) worth
persuing? It seems like it is, but wonder if the people creating public and
'for profit' backgammon software will adopt it?
[tyrade ends with ball bouncing not knowing what to do or if to do anything
next]
...Mark
damish@ll.mit.edu
|