Tracking Issue for const_ops

[AlexApps99]

Feature gate: #![feature(const_ops)]

This is a tracking issue for const_ops, which adds const versions of common numeric operations in std::ops.

Public API

• NonZero
  • BitOr
  • BitOrAssign
  • Div
  • Rem
• Wrapping
  • Shl
  • Shr
  • ShlAssign
  • ShrAssign
  • Add
  • AddAssign
  • Sub
  • SubAssign
  • Mul
  • MulAssign
  • Div
  • DivAssign
  • Rem
  • RemAssign
  • Not
  • BitXor
  • BitXorAssign
  • BitOr
  • BitOrAssign
  • BitAnd
  • BitAndAssign
  • Neg
• u*/i*
  • Add
  • Sub
  • Mul
  • Div
  • Rem
  • Neg
  • AddAssign
  • SubAssign
  • MulAssign
  • DivAssign
  • RemAssign
  • Not
  • BitAnd
  • BitOr
  • BitXor
  • Shl
  • Shr
  • BitAndAssign
  • BitOrAssign
  • BitXorAssign
  • ShlAssign
  • ShrAssign

Steps / History

• Implementation: Make most std::ops traits const on numeric types #89876.
• Tests
• Final comment period (FCP)
• Stabilization PR

Unresolved Questions

• None yet.

[AlexApps99]

Tests need to be implemented for this, I think the files library/core/tests/{nonzero,num/{int_macros,ops,uint_macros,wrapping}}.rs are good places to start.

I should be able to implement them in a week or two, once I can find the time to.

[kellerkindt]

FYI: Besides Wrapping there also exists a Saturating type

[benburk]

Any updates or blockers for this feature? Would love to be able to use these in my projects.

[AlexApps99]

I was originally planning to make tests, but I don't think that making any would be useful as it doesn't add any new functionality, and it relies on arithmetic already working in const.

Assuming tests aren't needed, the only other thing would be checking that it's ok to make this trait const. There was some debate when this PR was first made, I'm not sure what the policy is around const traits now.

[Mythra]

Hey, sorry to revive this, checking in on any potential updates/blockers. I tried converting some const functions which used .wrapping_mul, .wrapping_sub, etc. to use Wrapping types, and it pointed to this issue.

[AlexApps99]

Misclick! I can't think of any blockers, since const-ifying these operations doesn't necessitate tests when it uses already-const arithmetic.

It would be good to consult with somebody who can stabilize it, to get their thoughts

[TTWNO]

What needs to be done to stabilize this?

I'm looking to use this in tests. Willing to contribute tests, add docs if necessary, but let me know so we can land this.

[RalfJung]

This got entirely removed in #110393.

[scimind2460]

Shouldn't this be reopened as const_ops may be taken forward by the new syntax proposed by @fee1-dead?