Timothee Cour
Timothee Cour
PTAL, addressed comment; this PR is an improvement over status quo. > Well then, I propose using the system's source of randomness as a seed. If someone needs, for some...
> So this seems to be an argument of performance vs correctness. Either we use a source of cryptographic randomness as the seed, or a high-resolution monotonic timer. Is this...
> If there's a PR in the works for a proper fix, there's little point in a PR for an improper one, unless there's something time-critical going on. with this...
3 possible venues: * use atomics as done in std/oids which solves a very similar problem; can be augmented with thread-id + skipRandomNumbers to be robust to multiple threads returning...
@araq PTAL: * EDIT: depends on https://github.com/nim-lang/Nim/pull/18743 (std/cputicks; see corresponding RFC in https://github.com/nim-lang/RFCs/issues/411); it introduces a new module `std/cputicks` providing cpu instruction-level granularity, with much higher precision and lower overhead...
status as of https://github.com/nim-lang/Nim/pull/18050/commits/e425f0c082fce805a8dd67b727b52c06fd753a44: nim CI passes, 3 packages broken (last commit: fixup, containing: `# this is needed, see D20210519T200936 and D20210519T201000`)
> This does not actually fix #11225 from what I can tell. #11225 was already fixed (by using `bind` and making `bind` work properly), it's a separate topic under the...
PTAL, all important_packages now pass (each package that needed patching involved a tiny patch, typically adding a mixin, and the patch made sense in each case)
> Why not just allow general imports at local scope? done