[chore] Migrate from math/rand to math/rand/v2

[webstradev]

Description:

Migrating from math/rand to math/rand/v2.

Largely involves the following changes:

• Replaced any usage of math/rand.Read with crypto/rand.Read This was not carried over to the v2 API and the suggestion is to use the method from crypto.
• Removed any source creation / seeding -> No longer needed in math/rand/v2
• Applied renames for all outdated methods (lowercase n to uppercase N, and 31->32 and 63->64)

NOTE: There is one file where we now import both crypto/rand and math/rand/v2. I decided to alias both of them to avoid any confusion. So the imports in this package are

import (
    cryptoRand "crypto/rand"
    mathRand "math/rand"
)

I thought this would be best because having only one aliased could still be mistakenly read.

Link to tracking Issue: #34676

Testing:
Running make test

Documentation:

[webstradev]

still some work todo on this one will draft for now

[webstradev]

@mx-psi Two questions.

1. Seeding (like is done in countless places with the current time) is no longer necessary in v2, so I have removed it in those cases, but in the cases where we seed with a specific number I'm presuming we want a reproducible result, so in those cases I have gone with a different way of creating a source, because rand.NewSource does not exist in v2. So I've used rand.New(rand.NewPCG(seed, seed)) is that okay?
2. v2 removes the .Read implementation and suggests using crypto/rand.Read for true randomness, so I've done that for one case, but there are two cases where we seeded it with a specific seed (for reproducibility) as well. So not entirely sure how to proceed with those cases. I could alternatively leave those two packages out of this MR and that would leave only two packages on rand/v1

Thoughts?

[mx-psi]

@webstradev

So I've used rand.New(rand.NewPCG(seed, seed)) is that okay?

Yeah that seems okay, from https://go.dev/blog/chacha8rand#the-pcg-generator it seems like this is intended as the new default generator.

there are two cases where we seeded it with a specific seed (for reproducibility) as well. So not entirely sure how to proceed with those cases. I could alternatively leave those two packages out of this MR and that would leave only two packages on rand/v1

The godoc for rand.Read says:

If a deterministic source is required, use math/rand/v2.ChaCha8.Read.

Why don't we use this?

[webstradev]

@webstradev

So I've used rand.New(rand.NewPCG(seed, seed)) is that okay?

Yeah that seems okay, from https://go.dev/blog/chacha8rand#the-pcg-generator it seems like this is intended as the new default generator.

there are two cases where we seeded it with a specific seed (for reproducibility) as well. So not entirely sure how to proceed with those cases. I could alternatively leave those two packages out of this MR and that would leave only two packages on rand/v1

The godoc for rand.Read says:

If a deterministic source is required, use math/rand/v2.ChaCha8.Read.

Why don't we use this?

Yeah we can do this now, it was only added in 1.23 so when I was first looking at it it wasn't an option.
I'll see if I can sort this, however it does mean we need to update to 1.23, not sure if that is in scope for this PR? @mx-psi

[mx-psi]

Yeah we can do this now, it was only added in 1.23 so when I was first looking at it it wasn't an option.

Ah, I didn't realize this was Go 1.23 only 😞 I guess we can either:

• use crypto for it (not sure if this is feasible?) OR
• file an issue for tackling this once we support Go 1.23 and exclude these files from the linter for now

[webstradev]

Yeah we can do this now, it was only added in 1.23 so when I was first looking at it it wasn't an option.

Ah, I didn't realize this was Go 1.23 only 😞 I guess we can either:

• use crypto for it (not sure if this is feasible?) OR
• file an issue for tackling this once we support Go 1.23 and exclude these files from the linter for now

I think I'll go for the second option if that is okay. Do you usually mark a todo in code and reference an issue number as well

[mx-psi]

Do you usually mark a todo in code and reference an issue number as well

Yes, that's the usual way of doing it: put a TODO comment with a link to the issue

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 14 days.