Fruit 1.0 anniversary

[Ryan Benitez]

Today is 2014-03-17

Today is a good day for computer chess!

I downloaded senpai and give it a quick look (I will read it careful with time), I found a sense of familiarity and deja vu and I realized from where all this started: Fruit then Glaurung then Stockfish. Without your great engine even the engine I and other people work on would have not existed. IMHO you are one of the very few that really contributed to computer chess, one of the very few that opened a path that other people have followed.

So the news that you still have the will to dedicate time to this hobby is really a great news for all the community.

Sorry if now I ask a stupid technical question, but it is years that it bothers me. The random numbers of Polyglot book are really random or do you remember how you generated them? Eventually I'd like to use the original pseudo-random generator instead of copy the verbatim as are now:

https://github.com/mcostalba/Stockfish/ ... c/book.cpp

Thanks for your answer.

Glaurung used the coding style of Rock (Chess-64).

[hgm]

The random numbers of Polyglot book are really random or do you remember how you generated them? Eventually I'd like to use the original pseudo-random generator instead of copy the verbatim ...

Indeed, I second that. I would also very much like to remove that list from Polyglot and XBoard.

[Xann]

Hi Marco!

Needless to say, such high praise from the top programmer of the current era is a great honour.

Regarding the familiarity, it depends on the scanning level you applied: syntactic or semantic.

PolyGlot hash keys are random + filter. If that is your calling, there is some hope that an algorithm can be devised (estimated probability: 25%). The filter will take additional time though. Contact me if you are willing to go further.

[Xann]

Hi HGM!

This will not help you feel any better, but this table comes from development on Palm devices. There is no difference between "HDD" (protected RAM) and RAM. It therefore makes sense to put constant tables in the executable in order to minimise RAM usage, preferably under the 16KB limit of the older models.

[mcostalba]

Hi Marco!

Needless to say, such high praise from the top programmer of the current era is a great honour.

Regarding the familiarity, it depends on the scanning level you applied: syntactic or semantic.

PolyGlot hash keys are random + filter. If that is your calling, there is some hope that an algorithm can be devised (estimated probability: 25%). The filter will take additional time though. Contact me if you are willing to go further.

Thanks for your reply.

Familiarity is regarding choice of the names and factoring of the functions, so is a middle way between coding style (that is different) and semantic (that I still have to investigate).

I have sent you a PM regarding PolyGlot.

[PK]

There is one part of Fruit legacy I strongly dislike: refusal to use direction (vector) loops in eval_piece(). It was no problem in original Fruit, where these loops dealt only with pretty basic mobility. Current code, on the other hand, is a bit more complex, and therefore more repetitive. Trying to modify it would mean adding the same modification in 4 or 8 places, which seems painful.

So the question is: if I create a patch introducing loops and removing these repetitions, will it be accepted?

[Daniel Mehrmann]

There is one part of Fruit legacy I strongly dislike: refusal to use direction (vector) loops in eval_piece(). It was no problem in original Fruit, where these loops dealt only with pretty basic mobility. Current code, on the other hand, is a bit more complex, and therefore more repetitive. Trying to modify it would mean adding the same modification in 4 or 8 places, which seems painful.

So the question is: if I create a patch introducing loops and removing these repetitions, will it be accepted?

Yes, of course we'll look at this and accept if it fit. I agree with you that this special code is dangerous to do mistakes. It's complex and not clean looking anyway.

You can post the patch here or send me an e-mail. Best solution would be on github anyway.

[Jamal Bubker]

Thank you Fabien for reporting the birth of this new project !
Many thanks to Daniel et Ryan for creating and developping this new project !