I am trying to run Documenter.jl and the output results are different that the expected results due to a difference in sampling. I can reproduce a seeded result using 1 in both cases but different outputs! Any thoughts?
Why don’t you provide a seed argument to seed!()?
The documentation seems to suggest this https://docs.julialang.org/en/v1/stdlib/Random/#Random.seed!
Reseed the random number generator: rng
will give a reproducible sequence of numbers if and only if a seed
is provided.
Thanks! I generated the error using Random.seed!(1) and rand(2) to test. The actual issue is that the seed does not produce the same sequence in version 1.6.5 and version 1.7.1. What I had not realized is that both Atom and VSCode were on different releases. I know ther were some changes mad to move to Xoshiro 256+++ but I set is to Mersenne Twister for both tests. Really weird, no?
Vscode is not on any release. you either have julia in your path or it is specified in the vscode settings.
You may want to consider GitHub - JuliaRandom/StableRNGs.jl: A Julia RNG with stable streams