Re: [Bug-gnubg] Simpler PRNG

[Michael Petch]

On 2015-01-16 4:57 PM, Philippe Michel wrote:
> On Thu, 15 Jan 2015, Michael Petch wrote:
> 
>> I'm in agreement with Ian Shaw that a decrease in the options of random
>> number generators is probably more ideal. If others wish to weigh in on
>> this I am more than happy to listen.
> 
> Available RNGs are :
> 
> ansi
> bsd
> md5
> isaac 

These one I am in agreement with you and the assessment.

> bbs
> 
> The most profitable to drop. It causes a dependency on the gmp library.
> On the other hand, it brings a feature (cryptographically secure) that
> none of the other generators has. How important is it to have this ?
> 

This one is interesting. It is cryptographically secure so you can't
work backwards to find the state of the engine based on previously
produced value (and then predict the values in the future). libgmp is
not really a big issue IMHO. We also use it for Long seed support
(something that I put support in some time back). Interesting though it
appears there has been a GUI bug introduced that doesn't permit long
seed values to be entered in the GUI edit box (but is accepted at the
command line). I learned about this when Timothy Chow asked me if we
supported Long Seeds. So there is one individual who has a use for
libgmp to stay in. By default I have had libgmp as a dependency for the
MS Windows builds just to enable long seed support.

My opinion is that libgmp is already useful elsewhere. So the question
is does anyone use BBS or have I directed anyone to consider it. Only
two people comes to mind. Rec.games.backgammon most notable personality
in the past 15 years: Murat and a fellow named Roy Crabtree.

Whether that is enough to keep BBS - probably not. But I just wanted to
mention it for completeness.

> mersenne
> manual
> file
> random.org
> 

Agreed on these.

I also assume you'd be happy (like I am) to see all the dice
manipulation support be removed?

-- 
Michael Petch
GNU Backgammon Maintainer / Developer
OpenPGP FingerPrint=D81C 6A0D 987E 7DA5 3219 6715 466A 2ACE 5CAE 3304