Re: System.Random - bugs? or am I just confused?

I definitely agree that the StdGen fields should be unpacked. The slightly stricter evaluation of the randomRs generator seems like the right solution as well. -Edward On Thu, Jul 18, 2013 at 4:16 PM, Bas van Dijk wrote:

On 17 July 2013 19:00, Hans Georg Schaathun wrote:

...

Hi all,

I have been digging into the System.Random module lately. Initially, I was concerned about the split method, but trying to understand the code and algorithm, there are a couple of other issues that bother me. I hope this is the right forum to ask about them ...

Hi Hans, no comments on your issues yet. I just wanted to report an issue in random I discovered 3 months ago. This is the message I send to Ryan:

...

My colleague (Utkarsh Upadhyay) got a stack-overflow in a program he was working on. Thomas Schilling and me traced it down to randomRs being to lazy which causes the following to generate a stack-overflow exception:

g <- getStdGen randomRs (0,10::Int) g !! 1000000

The attached patch fixes the bug. I also included a patch to UNPACK the Int32 fields into the StdGen constructor.

I added the patches to the following gist:

https://gist.github.com/basvandijk/6032630

Cheers,

Bas

_______________________________________________ Libraries mailing list Libraries@haskell.org http://www.haskell.org/mailman/listinfo/libraries