Hi Alex.
Le mer. 20 mars 2019 à 01:32, Alex Herbert <[email protected]> a écrit :
>
>
>
> > On 19 Mar 2019, at 15:24, Gilles Sadowski <[email protected]> wrote:
> >>
> >> I am currently rerunning the dieharder test for the XorShift1024Star
> >> composites since that requires a little endian format on my machine. So
> >> far there are not as many failures when the byte order is reversed.
> >>
> >> Once that is done I think we can wrap this up by:
> >>
> >> - updating the stress test to support little/big endian format as input
> >> for the test suite
> >>
> >> - updating the stress test GeneratorsList to match the RandomSource enum
> >> order
> >>
> >> - merging the modified XorShift1024StarPhi generator
> >>
> >> - deprecating the XOR_SHIFT_1024_S enum in favour of XOR_SHIFT_1024_S_PHI
>
> I have updated the PR for the XorShift1024 variant to deprecate the old
> XOR_SHIFT_1024_S enum.
>
> Please take a look to see if the wording is OK:
>
> https://github.com/apache/commons-rng/pull/30/files#diff-630495e26d73fa22ada841dabde0ab47
>
> <https://github.com/apache/commons-rng/pull/30/files#diff-630495e26d73fa22ada841dabde0ab47>
How about:
/**
* @deprecated Since 1.3, where it is recommended to use {@code
XOR_SHIFT_1024_S_PHI},
* instead, due to its slightly better (more uniform) output.
* {@code XOR_SHIFT_1024_S} is still quite usable but both versions
share the exact same
* internal state, so that their outputs are correlated (and thus
should not be used together when
* independent sequences are assumed).
*/
?
>
> I have not deprecated the XorShift1024Star generator. That still works and is
> the base for the new one. Just deprecating the enum value serves as a warning
> for users to update, or use that generator with wisely.
+1
That's what the above text tries to convey.
Gilles
>
> Alex
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
