[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gnubg] Feature Request: Differential reporting of rollout resul
Re: [Bug-gnubg] Feature Request: Differential reporting of rollout results
Mon, 29 Jun 2009 11:17:45 -0400
> I find the argument that we should keep it because it is already
> very weak argument.
I should have phrased that differently. What I meant to say was a
variation on the old saying "If it isn't broken, don't fix it."
That's a cliche, but in many cases it is very true. I'm sure that
Christian and the other developers have many things on their to-do
lists which are far more pressing than removing things which do work.
The only problem I see with having the current roster of random
number generators is they add to the complexity of the interface. I'm
sorry, but I have no sympathy for people who find the interface too
complicated. The RNG selection is buried two levels deep in the
interface and the user who complains of complexity would never even
have to know that the choice exists. I think that Max's suggestion of
Mersenne Twister, manual dice, and random.org is actually a good one,
but still too severe. Keeping at the very least Random.org is
important. I would add to that list "read from file". Personally I
sometimes get my own random numbers from Random.org and have GnuBG
read the file.
Why should Random.org be included? It is a true random number
generator, not a pseudo random number generator. If you want to
appease critics, they can use that generator.
> Too many RNGs can be a bad thing.
Yes, we will have to agree to disagree on that one. I think having a
selection is a good thing.
> Perhaps RNG's should be removed 1 by 1, and see whether anyone
Again, why spend precious developer time fixing something which is
In the interest of compromise, I would not oppose the removal of the
Ansi, BSD, and either the ISSAC or MD5. I would have to research
those more to form a firm opinion. That would leave the Mersenne
Twister, the Blum, Blum, & Shub, Random.org, manual dice, and the
read from file. I am very opposed to removing all the generators
besides Mersenne Twister and manual dice. I think that with some
research into the various generators the list could be pared down but
not as severely as Ian and Max suggest.
Re: [Bug-gnubg] Feature Request: Differential reporting of rollout results,
Myshkin LeVine <=