Making an sgf better

Post Reply
ZeroTalent

Post by ZeroTalent »

Hi, I would like to know if you can somehow specify in your sgf that for example if Black plays anywere in the board exepting 3 or 4 key spots then White will respond with one particular move.
Thank you very much in advance,

ZeroTalent

{Posted by ZeroTalent}
unkx80

Post by unkx80 »

I think currently the SGF standard does not cater to this possibility, that is, not possible.

[quote]
Hi, I would like to know if you can somehow specify in your sgf that for example if Black plays anywere in the board exepting 3 or 4 key spots then White will respond with one particular move.
Thank you very much in advance,

ZeroTalent
[/quote]


{Posted by unkx80}
admin

Post by admin »

yeah, this isn't possible. i've often thought about special functionality like this, but it's difficult.

along these lines, i always wondered if there were some way to merge lines with different move order. sometimes A, B, C, D is the same as C, D, A, B. could the program automatically merge these lines so that the SGF files can be simpler? the thing is, sometimes the same result through a different order makes a difference, like with ko's. hard to make automatic. i'd be interested in people's thoughts.

adum

{Posted by admin}
Spica

Post by Spica »

[quote]
yeah, this isn't possible. i've often thought about special functionality like this, but it's difficult.

along these lines, i always wondered if there were some way to merge lines with different move order. sometimes A, B, C, D is the same as C, D, A, B. could the program automatically merge these lines so that the SGF files can be simpler? the thing is, sometimes the same result through a different order makes a difference, like with ko's. hard to make automatic. i'd be interested in people's thoughts.

adum
[/quote]

perhaps you could try a format such as move W[**] or W[xx] as meaning any move which isn't the first move in another variation from this node and a comma separated list for a smaller number of points which all require the same reply e.g W[aa,ab,ac].

as for your specific problem maybe something like SR[W[ab];B[ac]] could be recognised by the applet as 'from this node and lower nodes always reply to W[ab] with B[ac]' which may need to be cancelled later in the variation by ER[W[ab];B[ac]]. SR for standard response, ER for end response I was thinking.


{Posted by Spica}
Olivier

Post by Olivier »

[quote]along these lines, i always wondered if there were some way to merge lines with different move order. sometimes A, B, C, D is the same as C, D, A, B. could the program automatically merge these lines so that the SGF files can be simpler? [...]. i'd be interested in people's thoughts.[/quote]

I've already thought about that. I think that the tree structure in the SGF format is too rigid. Here is what I would suggest to do instead.

The file would consist of a list of nodes with a unique ID key (like in a database). Each node would have a list of the possible next moves'ID.

This scheme would allow to have more than one node pointing to a move, which would be very convenient for many problems where different situations come to the same answer.

Another feature could be to specify the next move together with an operation like "flip the board horizontally across column D". This would take care of symmetric problems that for the moment have only a line of play on one side.

A more complicated feature would be to allow the next move to depend on the one you come from. That might adress the ABCD problem of Adum.

The whole idea is very similar to the circulation inside a bunch of web pages related with links.

What do you people think about this?

-- Olivier

{Posted by Olivier}
Post Reply