Author: Vincent Diepeveen
Date: 09:39:26 01/28/02
Go up one level in this thread
On January 28, 2002 at 09:32:34, Severi Salminen wrote:
>Hi!
>
>Has there been any discussion about the next protocol version? What ideas has
>been presented?
>
>I have a few:
>
>Make the interface clearer:
>
>1. _Every single_ command should be first accepted by the engine before WB
>starts to use them - excluding "xboard" and moves. So if I don't want to see any
>"random", commands, I don't have to. This would make it easier to add new
>commands, debug, maintain...
I see 2 statements here
a) Every single_ command should be first accepted by the engine before WB
starts to use them
I disagree here. A protocol is a protocol. A command is INSIDE the
protocol set or it isn't.
b) i agree those types of commands are completely outdated.
>2. There should be only one ("force", or "pause", or "break" for example)
>command that engine needs to poll during the search. Now it is not clearly
it is even undocumented what force is doing. For example in crafty
force only works when it is in winboard mode. i regret that.
force in winboard is needed however to enter the moves played so far
from openings position.
>documented which commands I should poll during search and which commands follow
>a "force" command. And after this pausing command WB could send "Ping" to assure
>that engine has stopped thinking and send the actual command _after that_. Now
>it is also not clearly documented when "ping" might be sent.
ping is a good idea to add.
>3. Documentation should contain a list of "crucial" commands that an engine
>should support to play chess. This is for new programmers.
I agree here. More interesting is making 2 new command sets
a) default protocol with very simple commands and not hard to parse
stuff which is just enough to run online without getting loads of
control-c's and such (this gnuchess only thing we must be able to
throw overboard by now).
b) extended online freakset which has all the current features and is
not much different from what the protover2 is doing now
c) the user friendly set, which allows also an interface controlling
the engine more and giving more output for users.
Of course stupid commands like 'protover' we must skip too, it is
too freakish.
>4. Documentation should also include next info for _all_ commands:
>- when the command will be sent
>- in what state engine might be when it receives this command (if this command
>is sent during the search or not)
>- in what state the engine should stay after performing this command
>
>5. Maybe every command (that has been accepted, 1st point) should be confirmed
>by the engine after engine has really performed the necessary action. This would
>also help debugging. The command could be "done" but since it now is used at
>initialization, "ready" might be used.
>
>6. Maybe WB could send "setboard" after every move. There would be no need to
>program a move parses, engine wouldn't have to remember and save the game
>history. So allways when the board changes, "setboard" is sent. I can't think
>now any drawbacks in this but it would make things a lot easier: no need for
>"undo" and maybe "new" commands in analyze mode.
>
>7. The whole thing is to make WB simpler and more logical interface - including
>the docs. The protocol system is a great thing and it really allows the backward
>compatibility while still leaves room for development.
>
>8. Why there is (this is probably documented somewhere) so strong connection
>between GNUChess and WB? Why on earth are they distributed together?? Separating
>them would be beneficial for both.
>
>I have more ideas and maybe at some point even will do some programming but
>please comment these first. I'll post this also to WB forum. BTW: it is also
>possible that _I_ have not read the docs careful enough...
>
>Severi Salminen
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.