Scid 3.7 released

[GenoM]

great release but it is quite annoying to resize the windows every time you open Scid. Any plans to fix this?

regards,
Geno

If you save the layout (I suppose in docked windows mode), the layout #1 is automatically restored (menu options -> windows -> save layout).

Pascal Georges

Thanks, Pascal, it works

[Kurt Utzinger]

Just a great program I like to work with (3.7.1) under Linux Ubuntu.
Many thanks for all the work you are doing.
Greetings
Kurt

[Kirk]

Hi,

I released Scid 3.7 at http://scid.sourceforge.net/

There have been many changes (see Changelog file in source package), the most visible one being a new user interface (see http://scid.sourceforge.net/screenshots/docked.png).

Pascal Georges

Is there a SCID formatted large database out there?

[Denis P. Mendoza]

Hi,

I released Scid 3.7 at http://scid.sourceforge.net/

There have been many changes (see Changelog file in source package), the most visible one being a new user interface (see http://scid.sourceforge.net/screenshots/docked.png).

Pascal Georges

Is there a SCID formatted large database out there?

Try this link Cliff for a start:
http://sourceforge.net/project/showfile ... _id=223252

[Kirk]

Thanks Much Denis!!

[Andres Valverde]

Hi,

I released Scid 3.7 at http://scid.sourceforge.net/

There have been many changes (see Changelog file in source package), the most visible one being a new user interface (see http://scid.sourceforge.net/screenshots/docked.png).

Pascal Georges

Many thanks Pascal, I use Scid in a daily basis. Keep up the good work !

[hgm]

Hi,

I released Scid 3.7 at http://scid.sourceforge.net/

There have been many changes (see Changelog file in source package), the most visible one being a new user interface (see http://scid.sourceforge.net/screenshots/docked.png).

Pascal Georges

I have a request:

It seems that Scid is used as the major tool for tuning of Polyglot books. I have never engaged in book building or tuning, but it seems that one thing that people want is to set the weight of a move to zero, without actualy deleting the move. (I can't imagine why, but that's the way it is.)

Now Scid seems to never set the move weight to exactly zero, it uses something like 2^-16 for such moves, with as a consequence that sometimes such 'zero-weight' moves are played. I guess that the product of the number of Scid users and the number of games they play from their books is > 2^16, so there are frequent complaints about this.

Could you equip Scid with the possibility to set the move weight to exactly zero in a Polyglot book? Of course I am aware that some Polyglot versions have a bug that makes them crash if they encounter a zero weight, (this is just a friendly way of saying that they are actually sabotaged to do so ), so it would be OK to pop up a warning to the user that this would make the book unsuitable to run with every Polyglot version. But there are other programs that use Polyglot books (in particular WinBoard), so I could imagine that there are people that do not care at all if the book would make some Polyglot versions crash, and it would be really useful to them if they could set the weight to a mathematical zero.

[Eelco de Groot]

Yes, it would be great if that could be fixed!

I suppose that good bookmakers can work around this by only storing refutations in their .bin books, but not the actual move that is refuted. Because as Harm points out, as far as I know the probability of the move being playing can not be guaranteed to be zero in a polyglot .bin opening book. I just think it would help the overview you have, if you can see at a glance for which move there are still lines in the book, and this includes the moves that should not be played. In .ctg books these would be marked red, in Rebel books for instance it was also possible to mark moves as not to be played.

Maybe it was intentionally made this way not to give away too much information for which moves there are refutations in case you have a public book that is used for competition, and opponents also can see the book, in which case maybe you would want to be able to hide some of the moves? But you can still do that by deleting the move from book. I think it was a small Polyglot design oversight, more likely.

Maybe in the newer Polyglot versions from Fonzy Bluemers or from Michel van den Bergh there already have been attempts to fix it, but I am no expert on Polyglot, and I have not read anything about this particular issue.

Regards, Eelco

[hgm]

The probability of the move being played can be set to zero in a Polyglot book, but not if Polygot is using the book. When WinBoard is using the book there is no problem. It is just that there is no existing program that does set weights to zero.

I think the design error slipped in as follows: when Polyglot is used to create a book, it sets the weight as the number of games it found the move in, and moves that did not occur in any game simply will not be put in the book. So if during book ceation a zero weight occurs, it points ot to an error, and to gard against this error, an assert statement was included to make Polyglot terminate with an error message.

But the error is hat this assert is put at the point where Polyglot reads the book, not where it writes it. Which made no difference when Polyglot was the only creator and reader of such books, and therefore went unnoticed.

The assert on reading the book should have checked for the sum of all weights being different from zero which is a fatal condition, as it would leave the probability with which each of the moves should be played undetermined.

[F. Bluemers]

The probability of the move being played can be set to zero in a Polyglot book, but not if Polygot is using the book. When WinBoard is using the book there is no problem. It is just that there is no existing program that does set weights to zero.

I think the design error slipped in as follows: when Polyglot is used to create a book, it sets the weight as the number of games it found the move in, and moves that did not occur in any game simply will not be put in the book. So if during book ceation a zero weight occurs, it points ot to an error, and to gard against this error, an assert statement was included to make Polyglot terminate with an error message.

But the error is hat this assert is put at the point where Polyglot reads the book, not where it writes it. Which made no difference when Polyglot was the only creator and reader of such books, and therefore went unnoticed.

The assert on reading the book should have checked for the sum of all weights being different from zero which is a fatal condition, as it would leave the probability with which each of the moves should be played undetermined.

I'm not shure if scid itself can handle zero weight moves.
I created a book with a pg version that sets all weights to zero.