Re: Introducing Data Structure // A-Z in PropID's
- This letter is about
- syntax details
- last move
- collective game info property
Arno Hollosi wrote:
> We're trying to keep SGF simple and compatible :o)For this reasons I have suggested PI and M. You might
have questions about M here, but it is just one
property that respects coexistence of B and W in the
> CP[NihonKiin][SO:Kido vol.2][AN:KOBAYASHI Kouichi]IMO, this is the standard way SGF handles multiple values,
> [CO:KOBAYASHI Kouichi][AN.title_holding:Japanese Tengen]
so we should use the NN principle. We can then
decide whether we prefer NN[N1:bla][N2:blb][N3:blc][N4:bld]
or NN[bla blb blc bld] type of syntax. The first choice
can be used by databases, the second uses natural language.
I have no particular preferrence for either.
> Even this might cause problems with applications as they onlyIf some properties are unified, then minor problems occur anyway.
> expect one CP value.
By selecting a new unification property the trouble is smallest.
> CP[NihonKiinIMO, it is not necessary to invent yet another type of
> #SO:Kido vol.2
> #AN:KOBAYASHI Kouichi
> #AN.title_holding:Japanese Tengen
> #CO:KOBAYASHI Kouichi]
> RE[W+0.5I think that RE should not be used as a collective property
because it shall already contain the result type and comments
in details as in my proposal (like RE[W+F:B did not appear] ).
It is not clear to me what a Last_Move_Number_Property shall
be good for. Move numbers are implicitly given by the moves.
Passes give the end of a game phase. (I strongly oppose the
style of omitting final passes. Passes are moves like plays
are moves as well.) - Maybe a Last Move shall denote something
entirely different: The Last Move that has contested points.
Typically, in Asian literature it is this type of Last Move
that is particularly denoted. It is up to the annotator to
decide upon the proper move. However, this type of move does
not stop a game phase. It is the succession of passes that
stops a game phase. If it is this stop of a game phase (or
of the Alternation Phase in particular) that shall be given,
then the usage of B, W, or M with "pass" values is already
Also the remaining times are given by time properties rather
than by RE.
> PB[us:name in EnglishI would be happy with NN[us:Smith][jp:Japanese_name][cn:Chinese_name].
> #jp:name in Japanese
> #cn:name in Chinese]
However, I do not seriously argue about syntax details, I can live
with every syntax; powerful semantics is much more important.
With M a property with a composition of 3 values is suggested
(like M[1:W:cd] ). Since M is new anyway and the construction
of composition is as usual, the extension to 3 values within
one composition seems plausible to me. Otherwise one could split
M into 15 different properties, one for each move type, which
would be tedious.
PI for persons involved could be altered to the type
composed simpletext ':' simpletext
The first entry would specify the person type, one of:
Example: PI[RF:chief referee Kobayashi Koichi][RF:assistent
referee Cho Chikun][RF:assistent referee Rin Kaiho][CP:
Nihon Kiin][CP:Yomiuri][SO:Yomiuri][TK:Abe Yoshiteru]
The following alternative would allow miscallenous persons
PI[chief referee Kobayashi Koichi, assistent
referee Cho Chikun, assistent referee Rin Kaiho, copyright
Nihon Kiin and Yomiuri, source Yomiuri, time keeper Abe Yoshiteru,
famous kibitz Lee Chang-Ho]