Thanks for the update. I agree with deliberate approach.

Implementations of random int and math.random() < X predominately use the
upper bits. A different algorithm might make this distinction unnecessary,
but would it improve current behavior to bitflip the result? (or reverse groups
of bits
<https://gist.githubusercontent.com/anonymous/d09668b289fc6e4e4666/raw/e73b0db8624c0134a555b58e92e432c097e8a3c2/random.js>
)

Hard to determine goals of math.random since no documented specs or
requirements.

On Mon, Nov 23, 2015 at 12:40 AM, Yang Guo <[email protected]> wrote:

> Thanks for pointing that out. I am aware of the issue. However, I'd like
> to find a good alternative before replacing the existing one entirely. This
> is not a security issue, and rushing to implement a new PRNG could do more
> harm than keeping the old one until we found a good replacement.
>
> Yang
>
> On Mon, Nov 23, 2015, 04:30 Isaac Levy <[email protected]> wrote:
>
>> Did you see this comment
>> <https://codereview.chromium.org/1462293002#msg12> on your review?
>>
>> The upper bits of an MWC are low entropy, so the upper bits of
>>>
>>> your xor'd result I believe will be pretty low entropy too.
>>>
>>>
>> If the patch reduced entropy in the upper bits, then I don't think it
>> accomplished its goal. Might be worth another look -- it can cause subtle
>> issues <https://code.google.com/p/chromium/issues/detail?id=316359>,
>> especially when behavior varies across platform (webkit/jsc uses crypto
>> based generator)
>>
>> --
>> --
>> v8-dev mailing list
>> [email protected]
>> http://groups.google.com/group/v8-dev
>> ---
>> You received this message because you are subscribed to the Google Groups
>> "v8-dev" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [email protected].
>> For more options, visit https://groups.google.com/d/optout.
>>
> --
> --
> v8-dev mailing list
> [email protected]
> http://groups.google.com/group/v8-dev
> ---
> You received this message because you are subscribed to the Google Groups
> "v8-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.
>

-- 
-- 
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
--- 
You received this message because you are subscribed to the Google Groups 
"v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to