Add more details about const generics. #921

ehuss · 2021-01-06T18:20:36Z

This is an extension of #901 with some more details about const generics.

This is not intended to be exhaustive of everything regarding const generics, just to add a little bit more information that I'm aware of.

ehuss · 2021-01-06T18:21:39Z

Some notes about things I'm uncertain about:

I don't understand lazy_normalization_consts, so I don't know what needs to be said about "anonymous constants".
I did not mention that impl parameters must be constrained. I believe that goes for all parameters (type, lifetime, and const), and should probably be mentioned here, but I do not think there is anything special about const (though I may be wrong). I figure this could be added separately.
I don't think expr as TYPE < 0 needs to be called out as wrong (since < is interpreted as part of the type), but I am uncertain. I also don't think this is specific to const generics, but was something I noticed while reviewing it.
Does not discuss inference, since type inference is mostly undocumented.
Should it include the note about exhaustiveness? See the "exhaustiveness checking" section in stabilize #![feature(min_const_generics)] in 1.51 rust#79135. I think that applies to all generic parameters (including types and lifetimes), so I don't think it is specific to const generics. I'm also uncertain how to write it (is it just the impl params must be used in the impl type?).

Havvy

For exhaustiveness checking, it's basically saying that even if there are individual impls for all values of a const generic, that does not mean there is a blanket impl over all values of that const generic. Unlike types and lifetimes, there's not a constrained listing of all possible values that Rust knows about, so it's probably worth mentioning.

src/items/generics.md

Havvy · 2021-01-09T07:18:18Z

src/items/generics.md

+The generic parameters are in scope within the item definition where they are
+declared.
+
+[References], [raw pointers], [arrays], [slices][arrays], [tuples], and


This was here before, but I have no clue what this sentence is actually stating. Path syntax isn't mentioned at all up to this point.

Introduced in #258. If I had to guess, it is maybe trying to say is that the syntax for specifying arguments to those types does not use the same syntax that the other items (structs, fns, etc.) use. That is, instead of IDENT<…> they each have their own unique syntax for specifying arguments.

I agree, it seems a bit confusing. Any suggestions? Maybe @matthewjasper could clarify?

Yeah, I let my brain think about it overnight. I'm betting this statement was originally in paths.html where it makes sense and was moved here as part of a split out. We should probably bring most of GenericArgument into generics.html and then rename GenericArguments to GenericPathArguments or something. I'll open an issue for that one.

src/items/generics.md

ehuss · 2021-01-09T17:27:30Z

For exhaustiveness checking,

Sorry, I think my brain crossed its wires when I wrote that. The sentences towards the end were intended for the question about impl constrained parameters (the report_unsued_parameters check).

I took a stab at adding a small blurb about exhaustiveness checking, but I'm not comfortable with the wording. I suspect there are others who can word it more accurately.

Havvy

The blurb on exhaustive checking is a little clunky, but it gets the job done well enough. If somebody wants to word it better, they can always open a new PR.

ehuss · 2021-01-10T20:20:28Z

@Havvy I'm not sure if you had a chance to edit this, but I went ahead and pushed a small clarification on when const generics must be used.

Havvy

I have a more comprehensive explanation of a constraining generic parameter, but it has nothing to do with const parameters specifically, so I'm making that its own PR. The changes to generics.md you have are better than what I have, though.

Can you fix the file conflicts and we can merge this?

ehuss · 2021-01-11T06:16:13Z

Can you fix the file conflicts and we can merge this?

Done, I somehow missed the notification.

Update books ## nomicon 2 commits in a5a48441d411f61556b57d762b03d6874afe575d..a8584998eacdea7106a1dfafcbf6c1c06fcdf925 2020-12-06 10:39:41 +0900 to 2021-01-06 12:49:49 -0500 - Update vector code examples - Remove outdated information about `jemalloc` ## reference 13 commits in b278478b766178491a8b6f67afa4bcd6b64d977a..50af691f838937c300b47812d0507c6d88c14f97 2020-12-21 18:18:03 -0800 to 2021-01-12 21:19:20 -0800 - Update grammar for parser unification. (rust-lang/reference#927) - Define constraining an implementation (rust-lang/reference#928) - Document extra behavior of #[no_mangle] (rust-lang/reference#930) - Add a float examle without a `.`. (rust-lang/reference#929) - Add more details about const generics. (rust-lang/reference#921) - Fix footnotes. (rust-lang/reference#926) - Add "Logic errors" as behavior not considered unsafe (rust-lang/reference#919) - Update grammar for order of parameters/arguments. (rust-lang/reference#920) - Fix formatting in the tuple section (rust-lang/reference#923) - document const generics (rust-lang/reference#901) - Update mdbook (rust-lang/reference#918) - linkage.md: update link to FFI section of the Book. (rust-lang/reference#917) - Document array expression with a const. (rust-lang/reference#914) ## book 8 commits in 5bb44f8b5b0aa105c8b22602e9b18800484afa21..ac57a0ddd23d173b26731ccf939f3ba729753275 2020-12-18 20:07:31 -0500 to 2021-01-09 14:18:45 -0500 - Update version of mdbook we're testing with to 0.4.5 (rust-lang/book#2561) - Fix grammar in ch13-01-closures.md (rust-lang/book#2534) - Merge remote-tracking branch 'origin/pr/2527' - Clarify code example ch6.3 (rust-lang/book#2485) - Fix link added in rust-lang/book#2495 to be relative and at the bottom - Merge remote-tracking branch 'origin/pr/2495' - Update output to match the updated poem punctuation - Fix rust-lang/book#2539 - Remove fancy apostrophes from poem for Windows ## rust-by-example 3 commits in 1cce0737d6a7d3ceafb139b4a206861fb1dcb2ab..03e23af01f0b4f83a3a513da280e1ca92587f2ec 2020-12-21 17:36:29 -0300 to 2021-01-09 10:20:28 -0300 - Replace for loop with iteration (rust-lang/rust-by-example#1404) - Update mdbook (rust-lang/rust-by-example#1402) - Add note for match guards to include catch-all (rust-lang/rust-by-example#1401) ## embedded-book 1 commits in ba34b8a968f9531d38c4dc4411d5568b7c076bfe..ceec19e873be87c6ee5666b030c6bb612f889a96 2020-11-17 00:20:43 +0000 to 2021-01-03 13:13:10 +0000 - book.toml: add link to GitHub repo (rust-embedded/book#276)

Havvy requested changes Jan 9, 2021

View reviewed changes

ehuss force-pushed the const-generics-ext branch from f07b828 to 85a363c Compare January 9, 2021 17:31

Havvy mentioned this pull request Jan 10, 2021

Document Namespaces #924

Closed

Havvy self-requested a review January 10, 2021 04:43

Havvy approved these changes Jan 10, 2021

View reviewed changes

Havvy approved these changes Jan 11, 2021

View reviewed changes

ehuss added 3 commits January 10, 2021 22:11

Add more details about const generics.

ebb5552

Updates from review.

1c3659a

Clarify when const parameters need to be used.

7fd47ef

ehuss force-pushed the const-generics-ext branch from 7d86a9e to 7fd47ef Compare January 11, 2021 06:11

Havvy merged commit 3e93fea into rust-lang:master Jan 11, 2021

ehuss mentioned this pull request Jan 13, 2021

Update books rust-lang/rust#80973

Merged

Havvy mentioned this pull request Jan 13, 2021

Please link to reference #921 this week rust-lang/this-week-in-rust#1738

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add more details about const generics. #921

Add more details about const generics. #921

ehuss commented Jan 6, 2021

ehuss commented Jan 6, 2021 •

edited

Loading

Havvy left a comment

Havvy Jan 9, 2021

ehuss Jan 9, 2021

Havvy Jan 10, 2021

ehuss commented Jan 9, 2021

Havvy left a comment

ehuss commented Jan 10, 2021

Havvy left a comment

ehuss commented Jan 11, 2021

Add more details about const generics. #921

Add more details about const generics. #921

Conversation

ehuss commented Jan 6, 2021

ehuss commented Jan 6, 2021 • edited Loading

Havvy left a comment

Choose a reason for hiding this comment

Havvy Jan 9, 2021

Choose a reason for hiding this comment

ehuss Jan 9, 2021

Choose a reason for hiding this comment

Havvy Jan 10, 2021

Choose a reason for hiding this comment

ehuss commented Jan 9, 2021

Havvy left a comment

Choose a reason for hiding this comment

ehuss commented Jan 10, 2021

Havvy left a comment

Choose a reason for hiding this comment

ehuss commented Jan 11, 2021

ehuss commented Jan 6, 2021 •

edited

Loading