[Pachi] JSON fork

jonathan chetwynd j.chetwynd at btconnect.com
Mon May 30 16:06:01 CEST 2011


assigned to me,

patch due shortly

~:"

  	
Jonathan Chetwynd

j.chetwynd at btconnect.com
http://www.peepo.com/

+44 (0) 20 7978 1764


On 26 May 2011, at 14:50, jonathan chetwynd wrote:

> Petr,
> 
> excellent start.
> 
> can we use frame rather than interm, and move rather than final?
> frame has good resonances.
> 
> ---
> 
> please only include JSON and newline
> 
> ie
> remove /**/
> 
> and remove everything after JSON
> what is needed will be in JSON soon enough,
> and if wanted can easily be got by not setting reporting=json
> as per attached
> 
> ---
> 
> also please re-read json.org and try out jsonlint.org an excellent validator
> 
> I believe there is a typo viz:
> 
> {"interm": {"playouts": 10000, "best": {"F6": 0.581520}, "seq": ["F6","D6","E3","E4"], "can": [{"F6":0.582},{"E8":0.561},{"G8":0.537},{"E7":0.551}]}}
> 
> should read
> 
> {"interm":[{"playouts": 10000}, {"best": {"F6": 0.581520}},{"seq": ["F6","D6","E3","E4"]}, {"can": [{"F6":0.582},{"E8":0.561},{"G8":0.537},{"E7":0.551}]}]}
> 
> similarly for final or move.
> 
> as a minimum examples must validate, please advise me if you find validation errors in the glossary, this is critical.
> most unfortunately validation does almost nothing to confirm the internal logic, which can remain wildly mistaken.
> 
> use [] for arrays such as interm and final, {} for objects which can be comma separated as arrays
> 
> jsonbig must validate, but also territstatus looks interesting, this is one I had added to glossary as territory
> 
>  "territstatus": [{"A1":{"b":0.540,"w":0.459,"d":0.002}},{"B1":{"b":0.540}},{"C1":{"b":0.545}}
> 
> I'm not clear what the meaning of d difference is, so please edit glossary if d is useful.
> however if the sum is unity, it is clearly redundant to give all three values, and we also define the node order
> ie the order of values is always A1 to J9 in 9x9 game
> 
> {territstatus": [{0.540:2},{0.540:0},{0.545:0}]}
> 
> furthermore we need to agree on the range, 
> I had defined as -999 to +999 saves '0.' on every value see glossary
> however 0-499 white and 500-999 would save +/- sign for slight loss of accuracy and readability
> what do you think?
> 
> All games are stored in memory, so one has to condense where possible.
> there is more on this in other mail
> 
> best 
> 
> Jonathan
> 
> 
> 
>  	
> Jonathan Chetwynd
> 
> j.chetwynd at btconnect.com
> http://www.peepo.com/
> 
> +44 (0) 20 7978 1764
> 
> 
> {"interm": {"playouts": 10000, "best": {"F6": 0.581520}, "seq": ["F6","D6","E3","E4"], "can": [{"F6":0.582},{"E8":0.561},{"G8":0.537},{"E7":0.551}]}}
> {"interm": {"playouts": 20000, "best": {"F6": 0.581262}, "seq": ["F6","D6","E3","F5"], "can": [{"F6":0.581},{"E5":0.582},{"E7":0.571},{"F7":0.567}]}}
> {"interm": {"playouts": 30000, "best": {"F6": 0.581900}, "seq": ["F6","D6","E3","F5"], "can": [{"F6":0.582},{"E5":0.578},{"E7":0.567},{"F7":0.565}]}}
> {"interm": {"playouts": 40000, "best": {"F6": 0.579400}, "seq": ["F6","D6","E3","F5"], "can": [{"F6":0.579},{"E5":0.577},{"E7":0.567},{"F7":0.565}]}}
> {"interm": {"playouts": 50000, "best": {"F6": 0.579400}, "seq": ["F6","D6","E3","F5"], "can": [{"F6":0.579},{"E5":0.578},{"E7":0.567},{"F7":0.565}]}}
> {"interm": {"playouts": 60000, "best": {"F6": 0.577342}, "seq": ["F6","D6","E3","F5"], "can": [{"F6":0.577},{"E5":0.578},{"E7":0.566},{"F7":0.565}]}}
> {"interm": {"playouts": 70000, "best": {"E5": 0.581893}, "seq": ["E5","F7","G6","G7"], "can": [{"E5":0.582},{"F6":0.577},{"E7":0.566},{"F7":0.565}]}}
> {"interm": {"playouts": 80000, "best": {"E5": 0.584083}, "seq": ["E5","F7","G6","G7"], "can": [{"E5":0.584},{"F6":0.577},{"E7":0.566},{"F7":0.565}]}}
> {"final": {"playouts": 80001, "best": {"E5": 0.584083}, "seq": ["E5","F7","G6","G7"], "can": [{"E5":0.584},{"F6":0.577},{"E7":0.566},{"F7":0.565}]}}
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rover.ms.mff.cuni.cz/pipermail/pachi/attachments/20110530/abe08d6f/attachment.html>


More information about the Pachi mailing list