Infallible promotion #3027

RalfJung · 2020-11-30T15:41:54Z

This is the RFC-follow-up to rust-lang/lang-team#58.

matu3ba

Overall good read and strategy, but especially >> and << and bit-operators (^,&,|,!) are missing. They could require some more reasoning.

text/0000-infallible-promotion.md

matu3ba · 2020-12-01T18:33:44Z

text/0000-infallible-promotion.md

+If too much code is broken, various ways to weaken this proposal (at the expense of more technical debt, sometimes across several parts of the compiler) are [described blow][rationale-and-alternatives].
+
+The long-term plan is that such code can switch to [inline `const` expressions](2920-inline-const.md) instead.
+However, inline `const` expressions are still in the process of being implemented, and for now are specified to not support code that depends on generic parameters in the context, which is a loss of expressivity when compared with implicit promotion.


Does this mean, that this RFC is planned to be blocked on inline const expressions?

The pacing of the individual breaking changes will be up for the lang team to decide. Having a migration path may or may not make a difference when incurring a breaking change.

text/0000-infallible-promotion.md

RalfJung · 2020-12-02T13:07:59Z

but especially >> and << and bit-operators (^,&,|,!) are missing. They could require some more reasoning.

They are trivial, I just added them. What kinds of problems did you foresee here?

matu3ba · 2020-12-03T18:47:28Z

but especially >> and << and bit-operators (^,&,|,!) are missing. They could require some more reasoning.

They are trivial, I just added them. What kinds of problems did you foresee here?

I was assuming overflow just means overflow and not underflow, but this RFC specifies the meaning.
"shift value is unconditionally masked to be modulo N to ensure that the argument is always in range"

Thus should be fine.

RalfJung · 2020-12-03T19:24:27Z

Yeah, "overflow" includes all kinds of values that cannot be represented, no matter "in which direction they overflow".

bill-myers · 2020-12-04T04:35:35Z

How about changing constants so that if they have a #[panic] attribute they can have a "panic" value that results in a panic upon const use? (and no compiler error)

Then promotion would generate consts with the #[panic] attribute and everything should work fine without needing the complicated, inflexible and limited list of special cases in this RFC.

RalfJung · 2020-12-04T09:45:34Z

How about changing constants so that if they have a #[panic] attribute they can have a "panic" value that results in a panic upon const use?

That would need a lot of complicated machinery all over the compiler. Note that we need a "complicated, inflexible and limited list of special cases" anyway for promotion, e.g. to exclude interior mutability or types that drop. All this RFC is doing is restricting that list just a bit more to avoid having to deal with consts that can fail to evaluate without making compilation fail.

In other words, this RFC is a lot simpler than what you proposed. The patch to implement this RFC will be less than 20 lines, followed by a cleanup PR that can remove way more code (and way more subtle code) than what the RFC implementation added.

Also note that panics are not the only way consts can fail to evaluate. They can also cause Undefined Behavior or perform an operation that us not support by CTFE.

scottmcm · 2020-12-05T21:47:28Z

Thanks for the write-up, Ralf! I knew some of the various pieces here, but between this and its earlier MCP version still learned a bunch -- the details of how the checked operators are promoted was an interesting nuance, for example, and not something I'd seen documented elsewhere. It was great having it all in order to bring everything together.

Since it LGTM (and we asked you to write it),
@rfcbot fcp merge

rfcbot · 2020-12-05T21:47:28Z

Team member @scottmcm has proposed to merge this. The next step is review by the rest of the tagged team members:

No concerns currently listed.

Once a majority of reviewers approve (and at most 2 approvals are outstanding), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

rfcbot · 2020-12-18T04:47:28Z

🔔 This is now entering its final comment period, as per the review above. 🔔

RalfJung · 2020-12-20T12:25:09Z

A recent comment by @oli-obk reminded me of the subtle interaction of promotion and const-validity checks. I added some text explaining that to the RFC; please let me know if that makes sense.

text/0000-infallible-promotion.md

rfcbot · 2020-12-28T05:16:52Z

The final comment period, with a disposition to merge, as per the review above, is now complete.

As the automated representative of the governance process, I would like to thank the author for their work and everyone else who contributed.

The RFC will be merged soon.

RalfJung · 2021-01-02T13:56:38Z

I created a tracking issue at rust-lang/rust#80619.
Someone else will have to do the actual merging of the PR in this repo though. :)

RalfJung · 2021-01-09T13:43:37Z

@nikomatsakis @rust-lang/lang who's in charge of merging accepted RFCs? :)

KodrAus · 2021-01-13T04:05:39Z

I tried 😄

cc @Mark-Simulacrum

EDIT: un-cc @Mark-Simulacrum

KodrAus · 2021-01-13T04:08:02Z

Huzzah! The @rust-lang/lang team has decided to accept this RFC.

To track further discussion, subscribe to the tracking issue here:
rust-lang/rust#80619

…-obk avoid promoting division, modulo and indexing operations that could fail For division, `x / y` will still be promoted if `y` is a non-zero integer literal; however, `1/(1+1)` will not be promoted any more. While at it, also see if we can reject promoting floating-point arithmetic (which are [complicated](rust-lang/unsafe-code-guidelines#237) so maybe we should not promote them). This will need a crater run to see if there's code out there that relies on these things being promoted. If we can land this, promoteds in `fn`/`const fn` cannot fail to evaluate any more, which should let us do some simplifications in codegen/Miri! Cc rust-lang/rfcs#3027 Fixes rust-lang#61821 r? `@oli-obk`

make const_err a future incompat lint This is the first step for rust-lang#71800: make const_err a future-incompat lint. I also rewrote the const_err lint description as the old one seemed wrong. This has the unfortunate side-effect of making const-eval error even more verbose by making the const_err message longer without fixing the redundancy caused by additionally emitting an error on each use site of the constant. We cannot fix that redundancy until const_err is a *hard* error (at that point the error-on-use-site can be turned into a `delay_span_bug!` for uses of monomorphic consts, and into a nicely rendered error for [lazily / post-monomorhization evaluated] associated consts). ~~The one annoying effect of this PR is that `let _x = &(1/(1-1));` now also shows the future-incompat warning, even though of course we will *not* make this a hard error. We'll instead (hopefully) stop promoting it -- see rust-lang/rfcs#3027. The only way I see to avoid the future-incompat warning is to use a different lint for "failure to evaluate promoted".~~ Cc `@rust-lang/wg-const-eval`

RalfJung added 4 commits November 15, 2020 13:39

make implicit promotion infallible

dfcde45

add some info that was in the MCP but not here; link to MCP

5f08ece

incorporate some feedback

575bfac

use bulleted lists and add example for CheckedAdd promotion

4592892

matu3ba suggested changes Dec 1, 2020

View reviewed changes

RalfJung mentioned this pull request Dec 2, 2020

Warn when using [foo; 0] where foo is a non-const expression that evaluates to a Drop type rust-lang/rust#79580

Open

add MIR overflow check reference; mention bitwise and shift operators

49b6d56

scottmcm added the T-lang Relevant to the language team, which will review and decide on the RFC. label Dec 4, 2020

rfcbot added proposed-final-comment-period Currently awaiting signoff of all team members in order to enter the final comment period. disposition-merge This RFC is in PFCP or FCP with a disposition to merge it. labels Dec 5, 2020

rfcbot added final-comment-period Will be merged/postponed/closed in ~10 calendar days unless new substational objections are raised. and removed proposed-final-comment-period Currently awaiting signoff of all team members in order to enter the final comment period. labels Dec 18, 2020

add note on the interaction of promotion and const validity checks

508ae47

RalfJung mentioned this pull request Dec 20, 2020

WIP crater experiment: do not promote const fn calls implicitly ever rust-lang/rust#80243

Closed

meithecatte reviewed Dec 20, 2020

View reviewed changes

text/0000-infallible-promotion.md Outdated Show resolved Hide resolved

fix CheckedAdd example

4d747ee

RalfJung mentioned this pull request Dec 26, 2020

make const_err a future incompat lint rust-lang/rust#80394

Merged

rfcbot added finished-final-comment-period The final comment period is finished for this RFC. to-announce and removed final-comment-period Will be merged/postponed/closed in ~10 calendar days unless new substational objections are raised. labels Dec 28, 2020

RalfJung mentioned this pull request Jan 1, 2021

Tracking Issue for inline assembly (asm!) rust-lang/rust#72016

Closed

4 tasks

RalfJung mentioned this pull request Jan 2, 2021

Tracking Issue for infallible promotion rust-lang/rust#80619

Closed

2 tasks

KodrAus approved these changes Jan 13, 2021

View reviewed changes

RFC 3027: Infallible Promition

cb78c70

KodrAus merged commit 7711133 into rust-lang:master Jan 13, 2021

This comment has been minimized.

Sign in to view

bstrie mentioned this pull request Mar 15, 2021

const operands in asm! should use inline consts rather than const promotion rust-lang/rust#83169

Closed

RalfJung deleted the infallible-promotion branch September 30, 2022 14:08

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Infallible promotion #3027

Infallible promotion #3027

RalfJung commented Nov 30, 2020 •

edited by traviscross

Loading

matu3ba left a comment

matu3ba Dec 1, 2020

RalfJung Dec 2, 2020

RalfJung commented Dec 2, 2020

matu3ba commented Dec 3, 2020

RalfJung commented Dec 3, 2020

bill-myers commented Dec 4, 2020 •

edited

Loading

RalfJung commented Dec 4, 2020 •

edited

Loading

scottmcm commented Dec 5, 2020

rfcbot commented Dec 5, 2020 •

edited by joshtriplett

Loading

rfcbot commented Dec 18, 2020

RalfJung commented Dec 20, 2020

rfcbot commented Dec 28, 2020

RalfJung commented Jan 2, 2021

RalfJung commented Jan 9, 2021

KodrAus commented Jan 13, 2021 •

edited

Loading

KodrAus commented Jan 13, 2021

This comment has been minimized.

Infallible promotion #3027

Infallible promotion #3027

Conversation

RalfJung commented Nov 30, 2020 • edited by traviscross Loading

matu3ba left a comment

Choose a reason for hiding this comment

matu3ba Dec 1, 2020

Choose a reason for hiding this comment

RalfJung Dec 2, 2020

Choose a reason for hiding this comment

RalfJung commented Dec 2, 2020

matu3ba commented Dec 3, 2020

RalfJung commented Dec 3, 2020

bill-myers commented Dec 4, 2020 • edited Loading

RalfJung commented Dec 4, 2020 • edited Loading

scottmcm commented Dec 5, 2020

rfcbot commented Dec 5, 2020 • edited by joshtriplett Loading

rfcbot commented Dec 18, 2020

RalfJung commented Dec 20, 2020

rfcbot commented Dec 28, 2020

RalfJung commented Jan 2, 2021

RalfJung commented Jan 9, 2021

KodrAus commented Jan 13, 2021 • edited Loading

KodrAus commented Jan 13, 2021

This comment has been minimized.

RalfJung commented Nov 30, 2020 •

edited by traviscross

Loading

bill-myers commented Dec 4, 2020 •

edited

Loading

RalfJung commented Dec 4, 2020 •

edited

Loading

rfcbot commented Dec 5, 2020 •

edited by joshtriplett

Loading

KodrAus commented Jan 13, 2021 •

edited

Loading