2022年2月21日(月) 21:44 Tim Düsterhus <t...@bastelstu.be>: > Hi > > On 2/21/22 03:57, Go Kudo wrote: > > I am sorry for the delay in replying. > > Don't worry, there was a weekend inbetween and I totally understand that > one wants to take their weekend off. > > > Thank you for the clear explanation. > > It is true that the RFC in its current form lacks explanation. I'll try > to > > fix this first. > > Sounds good. > > > Also, as I look into other languages' implementations, I see the need to > > add some RNGs such as PCG. I will update the RFC to include these. > > I suggest you avoid "feature creep" within the RFC. Additional engines > can be added easily later on if the need arises. But for now it's more > important to get a reliable basis that one can build onto. > > Having a choice of a multitude of different engine just distracts from > that goal and can be confusing for the user. With xoshiro256** there's a > very good choice that already is part of the RFC, no need to have > something else that might or might not be slightly better in some case. > > Best regards > Tim Düsterhus >
Hi RFC has been updated. Is this up to the required standard? https://wiki.php.net/rfc/rng_extension I acknowledge that the previous RFC may have been difficult to discuss. If the problem has been solved, I would like to make another ML-wide announcement and wait for two weeks from there. I added PCG64 because according to the RNG experts, there seems to be a mild conflict between Xorshiro256 and PCG64. Also, as mentioned in the RFC, Rust and NumPy also implement PCG64. In order to verify the feasibility of PCG64, we created a PoC in C. So far, it seems to work fine. https://github.com/zeriyoshi/pcg64_example Regards, Go Kudo
