From: Tord Romstad <tord.romstad <at> gmail <dot> com>
Date: 26 jan 2007
Subject: Re: First draft of the Universal Shogi Interface (USI)

------=_Part_120979_21949998.1169803863531
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline


Hello, and thanks for the comments!

On 1/26/07, fat bold cyclop <fat.bold.cyclop <at> gmail <dot> com> wrote:
>
>
> Independently, some time ago, I came up with very similar approach to
> notation for shogi positions
> (http://www.javaforge.com/proj/wiki/displayPage.do?doc_id=27284 -- you
> have to join the project to be able to browse it's documentation, but
> the corresponding specification is almost the same as yours - except
> that it is based on EPD).
>
> But than I found shogiboard
> (http://www.shogi.net/nexus/shogiboard/index.html). There on the page
> you will notice endgame1.psn:
>
> [FEN
> "1nn6/2S+P1+L1P1/kgs4+R1/1lpP2pp1/KP2Sp3/l1P3P1p/+b1N1SP3/4G3R/3G3N1
> 2000110/4000000 s 55"]
>
> I don't know if he adopted some existing standard. I think the only flow
> of it is, I think, it insufficient for storing positions with more then
> 9 pieces in hand. I know it is probably an academic problem, but
> still...


In my opinion, this is a rather serious problem.  Positions with more than 9
pawns in hand are theoretically possible, even when the game starts from the
standard opening position.  I prefer a format which has exactly one
representation of every possible shogi position.

A brief summary of the feedback I have received so far:

The engine and author names should be moved to the "translation file",
because most engine and author names are likely to be Japanese.

The "go" command needs to support time controls with byoyomi.

Someone at Manabu's blog said (if I understood Google's translation
correctly) that it might be better to replace my representations of
positions and moves by the ones used by the kifu format.  I have no very
strong personal opinions here; I have no objections against switching to
kifu notation for positions and moves if most of the community thinks it
would be a good idea. Opinions for and against the two notations are
welcome.

It might be a good idea to inform the engine about the result at the end of
a game, in case the engine wants to do some kind of automatic learning based
on the result.

One person expressed an interest in support for shogi variants.  I'll
consider adding this, but because it is not likely that there will be many
GUIs or engines for non-standard variants, it doesn't have very high
priority.

Tord


--^----------------------------------------------------------------
This email was sent to: shogi-l <at> shogi <dot> net

EASY UNSUBSCRIBE click here: http://topica.com/u/?a2i6Ys.aBVYf3.c2hvZ2kt
Or send an email to: shogi-unsubscribe <at> topica <dot> com

For Topica's complete suite of email marketing solutions visit:
http://www.topica.com/?p=TEXFOOTER
--^----------------------------------------------------------------


------=_Part_120979_21949998.1169803863531
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

Hello, and thanks for the comments!<br><br><div><span class="gmail_quote">On 1/26/07, <b class="gmail_sendername">fat bold cyclop</b> &lt;<a href="mailto:fat.bold.cyclop <at> gmail <dot> com">fat <dot> bold <dot> cyclop@gmail <dot> com</a>&gt; wrote:</span>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br>Independently, some time ago, I came up with very similar approach to<br>notation for shogi positions
<br>(<a href="http://www.javaforge.com/proj/wiki/displayPage.do?doc_id=27284">http://www.javaforge.com/proj/wiki/displayPage.do?doc_id=27284</a> -- you<br>have to join the project to be able to browse it&#39;s documentation, but
<br>the corresponding specification is almost the same as yours - except<br>that it is based on EPD).<br><br>But than I found shogiboard<br>(<a href="http://www.shogi.net/nexus/shogiboard/index.html">http://www.shogi.net/nexus/shogiboard/index.html
</a>). There on the page<br>you will notice endgame1.psn:<br><br>[FEN<br>&quot;1nn6/2S+P1+L1P1/kgs4+R1/1lpP2pp1/KP2Sp3/l1P3P1p/+b1N1SP3/4G3R/3G3N1<br>2000110/4000000 s 55&quot;]<br><br>I don&#39;t know if he adopted some existing standard. I think the only flow
<br>of it is, I think, it insufficient for storing positions with more then<br>9 pieces in hand. I know it is probably an academic problem, but<br>still...</blockquote><div><br>In my opinion, this is a rather serious problem.&nbsp; Positions with more than 9 pawns in hand are theoretically possible, even when the game starts from the standard opening position.&nbsp; I prefer a format which has exactly one representation of every possible shogi position.
<br><br>A brief summary of the feedback I have received so far:<br><br>The engine and author names should be moved to the &quot;translation file&quot;, because most engine and author names are likely to be Japanese.<br><br>
The &quot;go&quot; command needs to support time controls with byoyomi.<br><br>Someone at Manabu&#39;s blog said (if I understood Google&#39;s translation correctly) that it might be better to replace my representations of positions and moves by the ones used by the kifu format.&nbsp; I have no very strong personal opinions here; I have no objections against switching to kifu notation for positions and moves if most of the community thinks it would be a good idea. Opinions for and against the two notations are welcome.
<br><br>It might be a good idea to inform the engine about the result at the end of a game, in case the engine wants to do some kind of automatic learning based on the result.<br><br>One person expressed an interest in support for shogi variants.&nbsp; I&#39;ll consider adding this, but because it is not likely that there will be many GUIs or engines for non-standard variants, it doesn&#39;t have very high priority.
<br></div></div><br>Tord

------=_Part_120979_21949998.1169803863531--