Explain the new valtree system for type level constants. #1097

oli-obk · 2021-03-29T10:19:20Z

Dependent on getting rust-lang/rust#83234 merged first.

cc @rust-lang/wg-const-eval

lcnr · 2021-03-29T10:45:14Z

src/const-eval.md

@@ -20,17 +20,44 @@ Additionally constant evaluation can be used to reduce the workload or binary
 size at runtime by precomputing complex operations at compiletime and only
 storing the result.

+All uses of constant evaluation can either be categorized as "influencing the type system"
+(array lengths, enum variant discriminants, const generic parameters), or as solely being


do discriminants influence the type system in any way?

I think ValTrees are fine here because enum variants have to be plain integers, but I don't think that this is strictly necessary

do discriminants influence the type system in any way?

as tags are computed from discriminants, yes, the tag is part of the representation.

yeah, how does the tag feel into the type system? It feels like I am missing something here

Well... You can technically transmute an enum into a layout equivalent datastructure and then use the discriminant to change an array length. But in a less whacky (and unsound) way, if you put a non-c-like enum into a const generic, the tag will end up in the ValTree and thus in the parameters of the const generic.

Enums are encoded by adding their tag as a field before all variant fields. Alternatively we could encode enums by having an Option<VariantIdx> on ValTree::Branches. There is no strong reason to prefer either, except that enums are rare right now, so we'd be pessimizing non-enum aggregates with an unused None field.

ah, yeah 🤔 if we encode the discriminant in the val tree we need to do it this way. Using the VariantIdx might be cleaner in the long run, but I don't really think it matters too much.

Maybe "affect the type system" is the wrong terminology... it's more about "does the value semantically matter to the compiler, or does the compiler just treat it as an opaque blobs (and the actual value is entirely irrelevant)". Discriminant values semantically matter -- the compiler computes on them to ensure uniqueness, determine layout niches, things like that.

Patterns, at least the subset of them that are considered for exhaustiveness checking, are another example of values that semantically matter, and that hence do/should use valtrees.

src/const-eval.md

lcnr · 2021-03-29T10:49:16Z

src/const-eval.md

+As a consequence, all decoding of `ValTree` must happen by matching on the type first and making decisions
+depending on that. The value itself gives no useful information without the type that belongs to it.
+One notable oddity is `&str` representation. There is no sized equivalent of it, so unlike slices we cannot
+choose to represent them as their sized variant (slices are represented as arrays). `&str` thus has


Is this a correct implication? I assume that we do so for perf reasons, I don't expect there to be any semantic reason we can't represent &str in the same way as &[u8].

well... there is that tiny difference described in the text, and I'm using it to justify the existance of a custom variant for &str. But you are right, the main reason is performance, but it didn't actually do much difference in my tests (at best going from an 18% regression to a 17% regression), so I may revert it in the future once I have a better handle on the perf issues

src/const-eval.md

camelid

This looks really good! I left some more stylistic comments (can't comment much on the content since I'm learning about valtrees as I'm reading this :). Totally fine to skip some of the suggestions if you prefer the way it is now :)

src/const-eval.md

jyn514 · 2021-04-27T17:12:07Z

I think @oli-obk is on break - @RalfJung would you be interested in following up on this? NBD if not, it can wait until they get back.

RalfJung · 2021-04-27T18:25:24Z

Given that rust-lang/rust#83234 didn't land yet, I think it might be a bit early to take this one over.

tshepang · 2022-08-09T20:24:30Z

closing as the PR it documents is also closed... feel free to re-open when that PR is revived

lcnr · 2022-08-10T09:38:46Z

valtrees have been merged in rust-lang/rust#96591

going to check if this document is still up to date

src/const-eval.md

JohnTitor · 2022-09-12T22:42:15Z

@oli-obk Friendly-ping, could you resolve merge conflicts and address the above comments? Once these are done, I think we could merge this!

oli-obk · 2022-09-13T11:02:31Z

done

src/const-eval.md

Co-authored-by: Noah Lev <[email protected]>

Update books ## nomicon 1 commits in d880e6ac2acf133dce640da24b9fb692844f02d4..f53bfa056929217870a5d2df1366d2e7ba35096d 2022-08-24 12:42:34 -0700 to 2022-09-05 07:19:02 -0700 - Small typo (rust-lang/nomicon#379) ## reference 9 commits in f62e93c28323ed9637d0a205a0c256498674a509..a7cdac33ca7356ad49d5c2b5e2c5010889b33eee 2022-08-28 10:01:28 -0700 to 2022-09-19 17:39:58 -0700 - Clarify wording for references. (rust-lang/reference#1223) - Update Unicode reference to match rustc implementation (rust-lang/reference#1271) - Add documentation for raw-dylib and link_ordinal (rust-lang/reference#1244) - Specify guarantees for repr(rust) structs (rust-lang/reference#1152) - Classify AsyncBlockExpression as ExpressionWithoutBlock (rust-lang/reference#1268) - Update closure-expr.md (rust-lang/reference#1269) - Clarify that 0 is a valid multiple of a type's alignment (rust-lang/reference#1260) - Remove `ne` from derive example (rust-lang/reference#1264) - Clarify reference on async blocks (rust-lang/reference#1262) ## book 6 commits in 0a5421ceb238357b3634fb75234eba4d1dad643c..f1e5ad844d0c61738006cdef26227beeb136948e 2022-08-28 19:51:04 -0400 to 2022-09-19 09:48:21 -0400 - Fix punctuation in ch05-02 - Ownership move chapter link fix - Wrong listing number - Reword text around box - `Box<T>` instead of "box" - Update Clippy output in Appendix D ## rust-by-example 2 commits in 03301f8ae55fa6f20f7ea152a517598e6db2cdb7..767a6bd9727a596d7cfdbaeee475e65b2670ea3a 2022-08-14 08:51:44 -0300 to 2022-09-14 09:17:18 -0300 - struct_visibility.md: Remove unneeded '#[allow(dead_code)]' (rust-lang/rust-by-example#1609) - Fix assorted typos (rust-lang/rust-by-example#1601) ## rustc-dev-guide 15 commits in 04892c1a6fc145602ac7367945fda9d4ee83c9fb..f587d6e7cddeaa3cf0a33ec1e368df1a408fa0aa 2022-08-29 20:07:51 +0200 to 2022-09-20 07:43:59 +0900 - Update stability guide to use CURRENT_RUSTC_VERSION (rust-lang/rustc-dev-guide#1468) - Add a note about building `rust-analyzer-proc-macro-srv` (rust-lang/rustc-dev-guide#1467) - Link from "implementing to new features" to mcp.md (rust-lang/rustc-dev-guide#1465) - remove stray ** - Explain the new valtree system for type level constants. (rust-lang/rustc-dev-guide#1097) - fix typos and formatting - Say "bootstrap" instead of "rustbuild"; the latter is not explained anywhere and is not much more clear. - Rewrite the section on passing flags to subcommands - Remove the diagram of all outputs generated by x.py - "symbol names" => ABI - Add symbol-addition to the how-to for new features (rust-lang/rustc-dev-guide#1457) - Fix typo (rust-lang/rustc-dev-guide#1459) - Document multipart_suggestion derive on SessionSubdiagnostic - Add reference for updating Windows PATH and fix typo - Update for removal of RLS (rust-lang/rustc-dev-guide#1450) ## embedded-book 1 commits in befe6840874311635c417cf731377f07234ee373..4ce51cb7441a6f02b5bf9b07b2eb755c21ab7954 2022-07-25 07:51:14 +0000 to 2022-09-15 08:53:09 +0000 - Create CITATION.bib (as per rust-embedded/book#327) (rust-embedded/book#329)

lcnr approved these changes Mar 29, 2021

View reviewed changes

camelid reviewed Mar 31, 2021

View reviewed changes

src/const-eval.md Outdated Show resolved Hide resolved

camelid reviewed Mar 31, 2021

View reviewed changes

src/const-eval.md Outdated Show resolved Hide resolved

camelid reviewed Apr 1, 2021

View reviewed changes

RalfJung reviewed Apr 3, 2021

View reviewed changes

src/const-eval.md Outdated Show resolved Hide resolved

RalfJung reviewed Apr 3, 2021

View reviewed changes

src/const-eval.md Outdated Show resolved Hide resolved

jyn514 added the S-waiting-on-author Status: this PR is waiting for additional action by the OP label Apr 27, 2021

jyn514 added S-blocked Status: this PR is blocked waiting for something and removed S-waiting-on-author Status: this PR is waiting for additional action by the OP labels Apr 27, 2021

oli-obk mentioned this pull request Jul 13, 2021

Handle non-integer const generic parameters in debuginfo type names. rust-lang/rust#87082

Merged

tshepang closed this Aug 9, 2022

lcnr reopened this Aug 10, 2022

lcnr reviewed Aug 10, 2022

View reviewed changes

src/const-eval.md Outdated Show resolved Hide resolved

src/const-eval.md Outdated Show resolved Hide resolved

jyn514 added S-waiting-on-author Status: this PR is waiting for additional action by the OP and removed S-blocked Status: this PR is blocked waiting for something labels Aug 10, 2022

oli-obk force-pushed the valtree branch from 3ba5e47 to 3d3cfbc Compare September 13, 2022 10:54

Explain the new valtree system for type level constants.

18521bb

oli-obk force-pushed the valtree branch from 3d3cfbc to 18521bb Compare September 13, 2022 11:02

lcnr reviewed Sep 13, 2022

View reviewed changes

src/const-eval.md Outdated Show resolved Hide resolved

src/const-eval.md Outdated Show resolved Hide resolved

lcnr and others added 2 commits September 13, 2022 14:31

Update src/const-eval.md

0d22497

Co-authored-by: Noah Lev <[email protected]>

Update src/const-eval.md

b984732

lcnr merged commit f1609a3 into master Sep 13, 2022

JohnTitor deleted the valtree branch September 13, 2022 12:48

ehuss mentioned this pull request Sep 21, 2022

Update books rust-lang/rust#102079

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Explain the new valtree system for type level constants. #1097

Explain the new valtree system for type level constants. #1097

oli-obk commented Mar 29, 2021

lcnr Mar 29, 2021 •

edited

Loading

oli-obk Mar 29, 2021

lcnr Mar 29, 2021

oli-obk Mar 29, 2021

lcnr Mar 29, 2021

RalfJung Apr 3, 2021 •

edited

Loading

lcnr Mar 29, 2021

oli-obk Mar 29, 2021

camelid left a comment

jyn514 commented Apr 27, 2021

RalfJung commented Apr 27, 2021

tshepang commented Aug 9, 2022

lcnr commented Aug 10, 2022

JohnTitor commented Sep 12, 2022

oli-obk commented Sep 13, 2022

Explain the new valtree system for type level constants. #1097

Explain the new valtree system for type level constants. #1097

Conversation

oli-obk commented Mar 29, 2021

lcnr Mar 29, 2021 • edited Loading

Choose a reason for hiding this comment

oli-obk Mar 29, 2021

Choose a reason for hiding this comment

lcnr Mar 29, 2021

Choose a reason for hiding this comment

oli-obk Mar 29, 2021

Choose a reason for hiding this comment

lcnr Mar 29, 2021

Choose a reason for hiding this comment

RalfJung Apr 3, 2021 • edited Loading

Choose a reason for hiding this comment

lcnr Mar 29, 2021

Choose a reason for hiding this comment

oli-obk Mar 29, 2021

Choose a reason for hiding this comment

camelid left a comment

Choose a reason for hiding this comment

jyn514 commented Apr 27, 2021

RalfJung commented Apr 27, 2021

tshepang commented Aug 9, 2022

lcnr commented Aug 10, 2022

JohnTitor commented Sep 12, 2022

oli-obk commented Sep 13, 2022

lcnr Mar 29, 2021 •

edited

Loading

RalfJung Apr 3, 2021 •

edited

Loading