bump `rand_core` to `0.10.0-rc.2` #95

baloo · 2025-11-05T19:41:04Z

No description provided.

baloo · 2025-11-05T19:41:20Z

codecov · 2025-11-05T19:54:09Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.46%. Comparing base (8b478bd) to head (54833c8).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #95      +/-   ##
==========================================
+ Coverage   97.44%   97.46%   +0.01%     
==========================================
  Files          14       14              
  Lines        1722     1734      +12     
==========================================
+ Hits         1678     1690      +12     
  Misses         44       44

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

tarcieri · 2025-11-05T20:05:47Z

Cargo.toml

 # need `crypto-bigint` with `alloc` to test `BoxedUint`
 crypto-bigint = { version = "0.7.0-pre.5", default-features = false, features = ["alloc"] }
-rand_chacha = "0.9"
+rand_chacha = "0.10.0-rc.1"


Perhaps switch to chacha20 v0.10.0-rc.4 here? I think the plan is to deprecate rand_chacha

I guess I can drop it altogether and use the one re-exported by rand?
https://docs.rs/rand/0.10.0-rc.1/rand/rngs/struct.ChaCha20Rng.html

rand_chacha::Chacha20Rng was Clone-able, chacha20::Chacha20Rng is not anymore. I've made a PR in rand-core to add a fork method in SeedableRng here:
rust-random/rand_core#17

This now depends on it.

Cargo.toml

baloo · 2025-11-06T03:08:39Z

src/multicore.rs

        loop {
            if let Some(result) = self.sieve.next() {
-                return Some(result);
+                return Some((self.rng.fork(), result));


the SieveIterator now injects a new forked Rng for each item it streams. This is the only way I've found to pass the rng to each predicate.

I think the weight of the operation is fairly minimal, but ... worth pointing it out.

Depends on the RNG, but probably negligible compared to testing a large number for primality. And definitely more correct than cloning the RNG.

Depends on the RNG,

To elaborate on that, cloning is probably cheaper than forking/splitting, but for RNGs that are counter based, splitting is cheap (essentially a clone and generate a nonce). That said, forking is the way to go.

Ideally you would rekey instead of generating a random nonce (ChaCha has 256-bit keys and 64-bit nonces for the variant we're using for RNGs, making collisions under random nonces too likely for comfort).

But the difference there is just extracting a bit more RNG output

@baloo can you just use *Rng::from_rng here for now with a TODO to use fork, so it isn't blocked on another rand_core release?

That's exactly what the default implementation of fork() does in rust-random/rand_core#17

@baloo can you just use *Rng::from_rng here for now with a TODO to use fork, so it isn't blocked on another rand_core release?

done.

Cargo.toml

fjarri · 2025-11-06T07:04:51Z

@tarcieri I assume crypto-bigint 0.7 will release with rand_core 0.10? Otherwise I'd rather stay on 0.9.

Don't worry about the build-docs and semver failures, I'll deal with it in another PR

Edit: actually, the doc-auto-cfg error comes from getrandom. Rust devs really screwed up a lot of builds here.

src/hazmat/sieve.rs

dvdplm · 2025-11-06T10:53:37Z

src/multicore.rs

        loop {
            if let Some(result) = self.sieve.next() {
-                return Some(result);
+                return Some((self.rng.fork(), result));


Depends on the RNG,

To elaborate on that, cloning is probably cheaper than forking/splitting, but for RNGs that are counter based, splitting is cheap (essentially a clone and generate a nonce). That said, forking is the way to go.

tarcieri · 2025-11-06T14:04:19Z

@fjarri yes, we're in the process of bumping everything to use rand_core v0.10, and can hopefully actually start cutting releases once it's been updated.

I've already cut a crypto-bigint v0.7.0-rc.10 prerelease that includes the upgrade.

tarcieri · 2025-11-06T20:43:19Z

The other failures look unrelated. Would it be possible to get this landed and get another prerelease out? Unfortunately this rand_core upgrade is pretty painful and I'm trying to get to baseline crate releases with initial support for everything

Cargo.toml

fjarri · 2025-11-06T21:23:44Z

src/multicore.rs

-        }))
+        Ok(iter
+            .par_bridge()
+            .find_map_any(|(mut rng, c)| if predicate(&mut rng, &c) { Some(c) } else { None }))


Hm, wait, how does that work? Why don't we need to clone the RNG anymore?

Clone on an RNG is not a really good idea as either forks of the clone will have a similar internal state and will yield the same bits.

Hence the idea to use SeedableRng::fork instead (which needs to take a &mut to mutate the internal state of the source RNG).

rand_chacha used to carry Clone: https://docs.rs/rand_chacha/latest/rand_chacha/struct.ChaCha20Core.html#impl-Clone-for-ChaCha20Core (rand_chacha is getting deprecated)
chacha20:rngs does not anymore: https://docs.rs/chacha20/0.10.0-rc.5/chacha20/struct.ChaCha20Rng.html

Here instead, the rng is injected by the iterator in each of the items (using fork).
It's a couple of lines down from this change.

(fork comes in the next rc of rand_core, I will send a followup to migrate)

`--html-in-header` apparently only works with the `--no-deps` option when `cargo doc` is invoked, which `cargo semver-checks` isn't doing This removes the rustdocflags overrides in `.cargo/config.toml` by deleting the file first.

tarcieri · 2025-11-06T21:54:10Z

@baloo this hack should fix the semver checks, but right now the build is breaking because of rand_core: #97

fjarri · 2025-11-06T23:57:43Z

That's weird, I'm getting a lot of compilation errors locally after this. Going to publish a new release after I figure it out.

fjarri · 2025-11-07T00:03:19Z

crypto-bigint in the main deps was still set to rc.5, that was the reason. Fixed by setting it to rc.10 as well.

tarcieri · 2025-11-07T01:12:54Z

@fjarri thanks for the release! Got everything updated

baloo marked this pull request as draft November 5, 2025 19:41

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch 2 times, most recently from d42c214 to 1accbdb Compare November 5, 2025 19:51

tarcieri reviewed Nov 5, 2025

View reviewed changes

tarcieri reviewed Nov 6, 2025

View reviewed changes

Cargo.toml Outdated Show resolved Hide resolved

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch from 1accbdb to badc078 Compare November 6, 2025 03:03

tarcieri reviewed Nov 6, 2025

View reviewed changes

Cargo.toml Outdated Show resolved Hide resolved

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch from badc078 to 54ba25d Compare November 6, 2025 03:06

baloo commented Nov 6, 2025

View reviewed changes

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch 2 times, most recently from 8c948c7 to 18e8361 Compare November 6, 2025 03:39

bump rand_core to 0.10.0-rc.2

7455a05

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch from 18e8361 to 7455a05 Compare November 6, 2025 03:45

baloo added 2 commits November 5, 2025 19:54

move rand_core to master

c56754d

fixup fmt

620edda

baloo marked this pull request as ready for review November 6, 2025 05:04

fjarri reviewed Nov 6, 2025

View reviewed changes

Cargo.toml Outdated Show resolved Hide resolved

dvdplm approved these changes Nov 6, 2025

View reviewed changes

tarcieri mentioned this pull request Nov 6, 2025

RC-9 Broke with newer version of crypto-primes RustCrypto/RSA#602

Closed

remove dependency on git release of rand_core

838d9c5

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch from 9a7e805 to b05b17a Compare November 6, 2025 18:47

adjust fmt

9030605

baloo force-pushed the baloo/rand_core/0.10.0-rc.2 branch from b05b17a to 9030605 Compare November 6, 2025 18:48

baloo mentioned this pull request Nov 6, 2025

chore: align rustfmt.toml with Cargo.toml #96

Merged

fjarri reviewed Nov 6, 2025

View reviewed changes

Cargo.toml Outdated Show resolved Hide resolved

fjarri reviewed Nov 6, 2025

View reviewed changes

tarcieri and others added 2 commits November 6, 2025 14:50

CI: fix cargo semver-checks

3efcd82

`--html-in-header` apparently only works with the `--no-deps` option when `cargo doc` is invoked, which `cargo semver-checks` isn't doing This removes the rustdocflags overrides in `.cargo/config.toml` by deleting the file first.

remove duplicate rand_core dep

f106651

bring tarcieri's semver-checks

54833c8

tarcieri mentioned this pull request Nov 6, 2025

CI: fix cargo semver-checks #97

Closed

fjarri merged commit d0f86f4 into entropyxyz:master Nov 6, 2025
12 checks passed

baloo deleted the baloo/rand_core/0.10.0-rc.2 branch November 7, 2025 01:24

baloo restored the baloo/rand_core/0.10.0-rc.2 branch November 7, 2025 01:24

bump rand_core to 0.10.0-rc.2 #95

bump rand_core to 0.10.0-rc.2 #95

Uh oh!

Conversation

baloo commented Nov 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

baloo commented Nov 5, 2025

Uh oh!

codecov bot commented Nov 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

baloo Nov 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

fjarri commented Nov 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tarcieri commented Nov 6, 2025

Uh oh!

tarcieri commented Nov 6, 2025

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tarcieri commented Nov 6, 2025

Uh oh!

Uh oh!

fjarri commented Nov 6, 2025

Uh oh!

fjarri commented Nov 7, 2025

Uh oh!

tarcieri commented Nov 7, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

bump `rand_core` to `0.10.0-rc.2` #95

bump `rand_core` to `0.10.0-rc.2` #95

baloo commented Nov 5, 2025 •

edited

Loading

codecov bot commented Nov 5, 2025 •

edited

Loading

baloo Nov 6, 2025 •

edited

Loading

fjarri commented Nov 6, 2025 •

edited

Loading