Author: Joachim Rang
Date: 07:25:53 12/07/02
Go up one level in this thread
On December 07, 2002 at 09:45:42, Miguel A. Ballicora wrote: >On December 07, 2002 at 09:29:36, David Rasmussen wrote: > >>On December 07, 2002 at 08:26:39, Steve Maughan wrote: >> >>>David, >>> >>>IMO it is definitely worth implementing but there are those who think it's far >>>inferior to Winboard >> >>Why do they think that? What is inferior about it? > >The whole design, IMHO. Too much control on the hands of the GUI, I do not like >it. UCI is great for an analysis engine, not good for a playing engine. >For instance, you have to do ponder only in the way that the GUI decides. You >cannot implement easily learning since the engine does not know when the game >finished. If you want an engine only for analysis, go for UCI, it supports >mutivariation mode, change easily the size of the tables etc. > >The problem is that the engine receives each time all the moves from the >beginning or a position. So, each move is like a new position. What do yo do >with the tables? do you erase them? you have to make a decision about this. >You can try to do clever things to try to guess if you are in a game or >analyzing a real new position, but I wonder if you can implement all that in 4-8 >hrs as Steve suggested. > >I have not implemented UCI because all the things above and because having >winboard I fail to see the advantage of another protocol. Then again, since I >did not do it, I might be missing something. > >YMMV, >Miguel > > > >> >>/David well "Too much control on the hands of the GUI" - this makes it inferior? From a customer perspective there can hardly be "too much control". I, like a customer, prefer UCI, becasue it's easier and quicker to change some values within the GUI instead of editing an Ini-File and restarting the engine. I like this kind of "too much control" because sometimes I have reason to decide that the games should be ended already with a score -5 or even lesser, or that the same book should be used.... If the user wishes so I see no problem. Of course since we are talking about free engines there is no base to "demand" the UCI-Protocol, but it would be nice ;-) I do not deny that there are limitations regarding pondering and learning and I'm not claiming that UCI is superior but nevertheless for me it's more comfortable.
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.