T-compiler Meeting Agenda 2023-12-21

Announcements

MCPs/FCPs

WG checkins

  • @_WG-llvm by @nagisa and @Nikita Popov (previous checkin):

    LLVM 18 is going to branch in about a month

Backport nominations

T-compiler stable / T-compiler beta

  • :beta: "Exhaustiveness: Improve complexity on some wide matches" rust#118796
    • PR is being reviewed, fixes a (P-high) exhaustiveness check rust#118437 (comment)
    • issue analysis and patch authored by @Nadrieril (big thanks!)
    • nominated by @apiraino, could be nice to have it backported to beta after merge (issue report affects a real-world crate)
  • No stable nominations for T-compiler this time.

T-types stable / T-types beta

  • No beta nominations for T-types this time.
  • No stable nominations for T-types this time.

PRs S-waiting-on-team

T-compiler

Issues of Note

Short Summary

P-critical

T-compiler

  • No P-critical issues for T-compiler this time.

T-types

  • No P-critical issues for T-types this time.

P-high regressions

P-high beta regressions

  • No new P-high regressions this time.

Unassigned P-high nightly regressions

  • No unassigned P-high nightly regressions this time.

Performance logs

triage logs for 2023-12-18

A lot of noise in the results this week; there was an lull in the noise
recently, so our auto-inferred noise threshold went down, and thus five PR's
were artificially flagged this week (and three supposed improvements were just
reverting to the mean). Beyond that, we had three nice improvements: the first
to debug builds in #117962 (by ceasing emission of expensive+unused
.debug_pubnames and .debug_pubtypes), a second to diesel and serde in
#119048 (by avoiding some unnecessary work), and a third to several benchmarks
in #117749 (by adding some caching of an internal compiler structure).

Triage done by @pnkfelix.
Revision range: 57010939..bf9229a2

Summary:

(instructions:u) mean range count
Regressions (primary) 0.3% [0.2%, 0.4%] 9
Regressions (secondary) 0.4% [0.2%, 0.9%] 21
Improvements (primary) -1.8% [-10.8%, -0.1%] 103
Improvements (secondary) -1.2% [-2.6%, -0.2%] 93
All (primary) -1.6% [-10.8%, 0.4%] 112

6 Regressions, 9 Improvements, 3 Mixed; 5 of them in rollups
67 artifact comparisons made in total

Regressions

rustc_passes: Enforce rustc::potential_query_instability lint #118870 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 13.5% [0.5%, 44.5%] 8
Regressions (secondary) - - 0
Improvements (primary) - - 0
Improvements (secondary) - - 0
All (primary) 13.5% [0.5%, 44.5%] 8
  • already marked as triaged by lqd; this PR did not regress things, the reported change is due to a periodic "slow s3 uploads on the collector", i.e. instrument noise/bias.

fix –dry-run when the change-id warning is printed #118789 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.9% [0.8%, 1.0%] 4
Regressions (secondary) 0.5% [0.3%, 0.8%] 18
Improvements (primary) - - 0
Improvements (secondary) - - 0
All (primary) 0.9% [0.8%, 1.0%] 4
  • already marked as triaged by lqd; this PR did not regress things. lqd hypothesized that the inferred noise threshold was lowered (due to a period of low noisiness?), and thus normal amounts of noise are tripping the threshold.

Rollup of 3 pull requests #118982 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.8% [0.8%, 0.9%] 4
Regressions (secondary) 0.4% [0.0%, 0.8%] 32
Improvements (primary) - - 0
Improvements (secondary) - - 0
All (primary) 0.8% [0.8%, 0.9%] 4
  • already marked as triaged by lqd. We think this is another case where the inferred noise threshold was low and we are seeing regression reports as the noise ramps up again.

Rollup of 5 pull requests #119002 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.2% [0.1%, 0.5%] 40
Regressions (secondary) 0.4% [0.1%, 0.7%] 36
Improvements (primary) -0.3% [-0.3%, -0.3%] 1
Improvements (secondary) -0.1% [-0.1%, -0.1%] 2
All (primary) 0.2% [-0.3%, 0.5%] 41
  • lqd identified PR #118396, "Collect lang items from AST, get rid of GenericBound::LangItemTrait" as the root cause for the regressions here.
  • not marking as triaged

More unescaping cleanups #118897 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.6% [0.1%, 0.8%] 5
Regressions (secondary) 0.4% [0.2%, 0.7%] 26
Improvements (primary) - - 0
Improvements (secondary) - - 0
All (primary) 0.6% [0.1%, 0.8%] 5
  • lqd theorized this is another instance of noise threshold wriggling
  • pnkfelix agrees and marked as triaged.

coverage: Regression test for assert!(!false) #119011 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.7% [0.7%, 0.8%] 4
Regressions (secondary) 0.5% [0.3%, 0.6%] 2
Improvements (primary) - - 0
Improvements (secondary) - - 0
All (primary) 0.7% [0.7%, 0.8%] 4
  • yet again, helloworld check is being flagged as regressing (in four scenarios), probably due to wriggling noise threshold.
  • pnkfelix marking as triaged

Improvements

Rollup of 8 pull requests #118838 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) - - 0
Improvements (secondary) -0.9% [-1.0%, -0.9%] 6
All (primary) - - 0

fix: stop emitting .debug_pubnames and .debug_pubtypes #117962 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -2.1% [-5.2%, -0.3%] 24
Improvements (secondary) -1.6% [-3.1%, -0.2%] 61
All (primary) -2.1% [-5.2%, -0.3%] 24
  • nice improvement to debug builds, especially for ripgrep and serde_derive

Unbreak non-unix non-windows bootstrap #118894 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -10.4% [-30.8%, -0.6%] 8
Improvements (secondary) - - 0
All (primary) -10.4% [-30.8%, -0.6%] 8
  • (don't get excited; this is just a return-to-normal after the artificial regression flagged with PR #118870 above)

cache param env canonicalization #117749 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -2.0% [-4.8%, -0.2%] 32
Improvements (secondary) -1.3% [-1.4%, -1.3%] 2
All (primary) -2.0% [-4.8%, -0.2%] 32
  • pretty solid improvement here to serde, diesel, bitmaps, hyper, image.

-Ztrait-solver=next to -Znext-solver #118937 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -0.9% [-1.0%, -0.9%] 4
Improvements (secondary) -0.5% [-0.8%, -0.3%] 21
All (primary) -0.9% [-1.0%, -0.9%] 4
  • (just noise)

Use the LLVM option NoTrapAfterNoreturn #110494 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -0.9% [-0.9%, -0.8%] 4
Improvements (secondary) -0.5% [-0.8%, -0.3%] 25
All (primary) -0.9% [-0.9%, -0.8%] 4
  • (just noise)

Add support for --env on tracked_env::var #118830 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -0.7% [-0.8%, -0.7%] 3
Improvements (secondary) -0.5% [-0.7%, -0.3%] 7
All (primary) -0.7% [-0.8%, -0.7%] 3
  • (just noise)

don't fold ParamEnv in register_predicate_obligation #119048 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -2.1% [-7.3%, -0.2%] 23
Improvements (secondary) -0.3% [-0.3%, -0.3%] 1
All (primary) -2.1% [-7.3%, -0.2%] 23
  • nice improvement to diesel and serde.

resolve: Replace visibility table in resolver outputs with query feeding #118657 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) - - 0
Improvements (primary) -0.3% [-0.5%, -0.1%] 70
Improvements (secondary) -0.6% [-1.2%, -0.1%] 42
All (primary) -0.3% [-0.5%, -0.1%] 70

Mixed

codegen_llvm: set DW_AT_accessibility #115165 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.2% [0.2%, 0.3%] 4
Regressions (secondary) - - 0
Improvements (primary) -0.7% [-0.7%, -0.6%] 4
Improvements (secondary) -0.4% [-0.7%, -0.2%] 22
All (primary) -0.2% [-0.7%, 0.3%] 8
  • the improvements are helloworld noise; the regressions are minor.

Rollup of 6 pull requests #119009 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) 0.2% [0.2%, 0.2%] 1
Regressions (secondary) 0.4% [0.2%, 1.2%] 8
Improvements (primary) -1.3% [-1.9%, -0.9%] 8
Improvements (secondary) - - 0
All (primary) -1.1% [-1.9%, 0.2%] 9
  • The single tiny regression to exa doc full does not warrant spending time trying to tease apart this rollup, IMO.
  • marked as triaged

Rollup of 5 pull requests #119069 (Comparison Link)

(instructions:u) mean range count
Regressions (primary) - - 0
Regressions (secondary) 0.3% [0.3%, 0.3%] 2
Improvements (primary) -0.2% [-0.3%, -0.1%] 12
Improvements (secondary) -0.3% [-0.5%, -0.2%] 13
All (primary) -0.2% [-0.3%, -0.1%] 12
  • The two minor regressions to secondary benchmarks do not warrant spending time trying to tease apart this rollup, IMO.
  • marked as triaged

Nominated Issues

T-compiler

  • No I-compiler-nominated issues this time.

RFC

  • No I-compiler-nominated RFCs this time.

Oldest PRs waiting for review

T-compiler

  • "tidy watcher" rust#114209 (last review activity: 2 months ago)
    • anyone wants to adopt this puppet? Seems it's bouncing a bit around
  • "Try Carbon's new hashing strategy in the compiler" rust#117079 (last review activity: about 58 days ago)
    • cc: @bjorn3 (IIUC)
  • "mark ty::Const::Error when meet unsupport ty for const generic params" rust#117176 (last review activity: about 56 days ago)
    • cc: @michael goulet (compiler-errors) (for some questions but review is assigned to @Wesley Wiser)
  • "Make #![allow_internal_unstable(..)] work with stmt_expr_attributes" rust#117420 (last review activity: about 51 days ago)
    • cc: @wesley wiser

Next week's WG checkins

  • @WG-mir-opt by @oli (on leave, other WG member?)
  • @_T-rust-analyzer by @Lukas Wirth

Next meetings' agenda draft: hackmd link

Select a repo