[mercury-users] Converting time_t to int using common library?

ok ok at cs.otago.ac.nz
Thu Feb 22 14:20:06 AEDT 2007


On 20 Feb 2007, at 6:47 pm, Mark Brown wrote:
> 	All random number generators have their weak points.
>
> 	Every RNG has its deficiencies. No RNG is appropriate for all tasks.
>
> This is why I've been objecting all along to the idea of putting  
> "the best"
> RNG in the standard library, or of giving our endorsement to any  
> particular
> implementation, or indeed of implying that there even exists such a  
> thing
> as the best RNG.

It would simply be morally wrong to provide a random number generator  
that was KNOWN
to be bad.  Disclaimers in various places never seem to work; there  
will always be someone
who decides to trust the implementors anyway.

It would also be wrong to provide no random number generator.   
Anything the Mercury
team would seriously consider providing would be much better than  
anything a (random-)
naive user would come up with themselves.  At best, users will say  
nasty things about the
Mercury team and try to adapt the first algorithm in the first book  
(or web page) they find that
has one.

So at least one good (but not 'best') RNG should be provided.  The  
only way to persuade
people that there is a choice for them to make is to GIVE them a  
choice.  This means an
interface and two implementations.  Making the choice should be 1  
line of code, but there
should be that 1 line, not least because in a few years it might  
prove advisable to replace
one or more of the implementations with something better.

--------------------------------------------------------------------------
mercury-users mailing list
Post messages to:       mercury-users at csse.unimelb.edu.au
Administrative Queries: owner-mercury-users at csse.unimelb.edu.au
Subscriptions:          mercury-users-request at csse.unimelb.edu.au
--------------------------------------------------------------------------



More information about the users mailing list