[core] make static vars const for enforced future thread safety

[ferdymercury]

This Pull request:

Changes or fixes:

Fixes #18751

This applies the suggestion of Dr15Jones

Checklist:

• tested changes locally
• updated the docs (if necessary)

[guitargeek]

It's possible to determine the lengths at compile time. Like this, we don't need to rely on static variables here at all.

[guitargeek]

Maybe I'm missing something, but how does adding const make this thread safe? if two threads call the function at the same time for the first time, would initLengthsVector() not be called concurrently still? Anyway with constexpr there won't be a problem.

edit: actually, static const is safe: https://stackoverflow.com/questions/8102125/is-local-static-variable-initialization-thread-safe-in-c11 Well, so I guess the constexpr version is just an optimization :)

[pcanal]

Maybe I'm missing something, but how does adding const make this thread safe? if two threads call the function at the same time for the first time, would initLengthsVector() not be called concurrently still

Just the static make the initialization thread safe. The C++ standard guarantees that a function static variable initialization is thread safe (and this is the reason for initialization pattern using a lambda function). The const only adds a guarantee that there will be no inadvertent updates to the static values (those would of course not be inherently thread-safe and would need their own locking mechanism)

[pcanal]

In practice this is a NFC change since the variable were never modified but a good protection against inadvertent changes. Thanks.

[ferdymercury]

In practice this is a NFC change since the variable were never modified but a good protection against inadvertent changes. Thanks.

Yep, i tried to reflect that in the PR title

[github-actions]

Test Results

    19 files      19 suites   3d 16h 56m 45s ⏱️
 2 746 tests  2 746 ✅ 0 💤 0 ❌
50 731 runs  50 731 ✅ 0 💤 0 ❌

Results for commit 1f02576.

[pcanal]

See #18757