T-compiler Meeting Agenda 2022-08-04

Announcements

Types Team: Planning/Deep-Dive meeting at time:2022-08-05T09:00:00-04:00
Next Thursday August, 11th release of Rust stable 1.63
Reminder: if you see a PR/issue that seems like there might be legal implications due to copyright/IP/etc, please let the Core team know (or at least message @pnkfelix or @Wesley Wiser so we can pass it along).

Other WG meetings

wg-rls-2.0 weekly sync-up at time:2022-08-08T11:00:00-04:00

MCPs/FCPs

New MCPs (take a look, see if you like them!)
- “Use RangeInclusive in SpanData instead of lo/hi” compiler-team#534
- “Rewrite compiletest out of tree” compiler-team#536
Old MCPs (not seconded, take a look)
- “Arbitrary annotations in compiletest” compiler-team#513 (last review activity: 3 months ago)
- “Add support for the LoongArch architecture” compiler-team#518 (last review activity: about 48 days ago)
- “Add #[alias] attribute to allow symbol aliasing” compiler-team#526 (last review activity: about 13 days ago)
Pending FCP requests (check your boxes!)
- “[RFC] Support .comment section like GCC/Clang (!llvm.ident)” rust#97550
- “Update cc-rs to 1.0.73 for compiler + bootstrap” rust#99477
Things in FCP (make sure you’re good with it)
- “Debuggable Macro Expansions” compiler-team#386
- “Add import_name_type parameter to #[link]” compiler-team#525
- “Re-add most early unstable syntax gates as future-compat lints” compiler-team#535
- “Introduce core::marker::Tuple to properly type-check extern "rust-call" calls” compiler-team#537
- “Increase the minimum linux-gnu versions” rust#95026
Accepted MCPs
- “Removing codegen logic for nvptx-nvidia-cuda (32-bit target)” compiler-team#496
- “Introduce -Z flag to control how proc-macros are run” compiler-team#528
- “move hir typeck into separate crate” compiler-team#529
- “Create #[rustc_on_type_error] mimicking #[rustc_on_unimplemented] for E0308 customization” compiler-team#530
- “Split Predicate into Goal and Clause” compiler-team#531
- “Deref is not a projection” compiler-team#532
- “Remove HashStable impl for collection types with unstable iteration order” compiler-team#533
Finalized FCPs (disposition merge)
- “Stabilize -Zgcc-ld=lld as -Clink-self-contained=linker -Clinker-flavor=gcc-lld” compiler-team#510
- “Remove migrate borrowck mode” rust#95565
- “Modify MIR building to drop repeat expressions with length zero” rust#95953
- “Lang: Stabilize usage of rustc_nonnull_optimization_guaranteed on -1” rust#97122
- “Remove a back-compat hack on lazy TAIT” rust#97346
- “Make outlives::{components,verify} agree” rust#97406
- “make cenum_impl_drop_cast deny-by-default” rust#97652
- “make const_err show up in future breakage reports” rust#97743
- “lub: don’t bail out due to empty binders” rust#97867
- “allow unions with mutable references and tuples of allowed types” rust#97995
- “do not mark interior mutable shared refs as dereferenceable” rust#98017
- “session: stabilize split debuginfo on linux” rust#98051

WG checkins

@_WG-async-foundations by @nikomatsakis and @tmandry (previous checkin):
Checkin text
@_WG-traits (Generic associated types initiative) by @Jack Huey (previous checkin):
No new updates - stabilization PR is FCP-merged for lang+types

Backport nominations

T-compiler beta / T-compiler stable

:beta: “allow numbers with provenance within CTFE execution” rust#99965
- There are some undocumented UB behaviours or (iiuc) indeed documented but not enough well communicated (comment)
- This beta’s backport goal seems to revert some of rust#99965 and allow time for a discussion on rust#99923 to happen
:beta: “Fix backwards-compatibility check for tests with +whole-archive” rust#100068
- Fixes rust#100066
- introduced by mistake in 1.61 (rust#95606) (comment)
:stable: “Fix backwards-compatibility check for tests with +whole-archive” rust#100068
- stable nomination only in case a 1.63.1 will be planned (comment)

T-rustdoc beta / T-rustdoc stable

No backport nominations for T-rustdoc this time.

:back: / :shrug: / :hand:

PRs S-waiting-on-team

T-compiler

“Tracking issue for dyn upcasting coercion” rust#65991
- Unsure about the status, waiting on T-lang?

Oldest PRs waiting for review

T-compiler

“Suggest using an appropriate keyword for struct and enum” rust#94996 (last review activity: 4 months ago)
- cc: @Esteban Küber
“Implement special-cased projection error message for some common traits” rust#98863 (last review activity: about 31 days ago)
- rustbot autoassigned @Esteban Küber (reroll?)
“Track derive attrs for more accurate suggestion” rust#98965 (last review activity: about 29 days ago)
- perhaps waiting on @Vadim Petrochenkov

Issues of Note

Short Summary

P-critical

T-compiler

“Wrong cast of u16 to usize on aarch64” rust#97463
- PR rust#97800 from Felix was reviewed and r’ed from Wesley

T-rustdoc

No P-critical issues for T-rustdoc this time.

P-high regressions

P-high beta regressions

“Source of lifetime coercion is not reported starting in 1.63” rust#99256
- Assigned P-high (comment) but that’s open to discussion

Unassigned P-high nightly regressions

“for<'a> &'a T: 'a and closures regressed” rust#98437
- @Jack Huey adds a comment from T-types (see comment)
- Close this issue? Remove the regression label?

Performance logs

triage logs for 2022-08-02

A good week! We had one regression from #99123 that could use further study, but we had a huge swath of improvements. The downward trend of the performance graph summary is striking. Perhaps best of all: Only one rollup PR impacted performance this week which makes things so much easier to evaluate! Great work everyone!

Triage done by @pnkfelix. Revision range: 50166d5e..792bc5a0

Summary:

	mean	max	count
Regressions (primary)	0.4%	0.7%	5
Regressions (secondary)	0.6%	1.1%	11
Improvements (primary)	-0.8%	-3.3%	140
Improvements (secondary)	-0.9%	-11.0%	95
All (primary)	-0.8%	-3.3%	145

1 Regressions, 4 Improvements, 4 Mixed; 1 of them in rollups 30 Untriaged Pull Requests 44 artifact comparisons made in total

Regressions

proc_macro: use crossbeam channels for the proc_macro cross-thread bridge #99123 (Comparison Link)

	mean	max	count
Regressions (primary)	0.9%	1.4%	11
Regressions (secondary)	N/A	N/A	0
Improvements (primary)	N/A	N/A	0
Improvements (secondary)	N/A	N/A	0
All (primary)	0.9%	1.4%	11

all 11 regressions are to various profiles+scenarios for html5ever
@nnethercote and @eddyb are discussing the regression to html5ever on the PR (#99123)
not marking as triaged, not yet at least; lets try and see if eddyb’s suggested mitigation actually helps here

Improvements

Rollup of 7 pull requests #99816 (Comparison Link)

	mean	max	count
Regressions (primary)	N/A	N/A	0
Regressions (secondary)	1.0%	1.0%	1
Improvements (primary)	-0.2%	-0.2%	1
Improvements (secondary)	-0.5%	-1.6%	10
All (primary)	-0.2%	-0.2%	1

primary improvement was hyper check incr-unchanged; secondaries were coercions debug full, and 9 variations on extern. Secondary regression was deep-vector opt incr-patched: println.

anonymize all bound vars, not just regions #99730 (Comparison Link)

	mean	max	count
Regressions (primary)	N/A	N/A	0
Regressions (secondary)	N/A	N/A	0
Improvements (primary)	-0.4%	-0.6%	56
Improvements (secondary)	-0.4%	-0.9%	36
All (primary)	-0.4%	-0.6%	56

As is, this PR registered compilation time improvements
However, @simulacrum noted that it was associated with a nearly 2% regression to bootstrap times.

Limit symbols exported from proc macros #99944 (Comparison Link)

	mean	max	count
Regressions (primary)	N/A	N/A	0
Regressions (secondary)	N/A	N/A	0
Improvements (primary)	-0.6%	-2.9%	41
Improvements (secondary)	-2.4%	-10.7%	11
All (primary)	-0.6%	-2.9%	41

wow, this was a big win! serde_derive incr-unchanged builds up to 3% faster; ripgrep check is up to 1.75% faster, hyper and webrender check 1% faster.

Lexer improvements #99884 (Comparison Link)

	mean	max	count
Regressions (primary)	N/A	N/A	0
Regressions (secondary)	N/A	N/A	0
Improvements (primary)	-0.3%	-0.6%	42
Improvements (secondary)	-0.5%	-1.2%	48
All (primary)	-0.3%	-0.6%	42

This was a very broad win, where a huge number of benchmarks all improved by a relatively small amount.
I’m impressed by the inventiveness shown during PR review.

Mixed

Remove remaining uses of box syntax from librustdoc #99577 (Comparison Link)

	mean	max	count
Regressions (primary)	N/A	N/A	0
Regressions (secondary)	0.3%	0.3%	4
Improvements (primary)	-0.3%	-0.3%	2
Improvements (secondary)	N/A	N/A	0
All (primary)	-0.3%	-0.3%	2

as you might expect, the limited number of performance changes here were to doc benchmarks (specifically serde and libc doc).

Remove TreeAndSpacing. #99887 (Comparison Link)

	mean	max	count
Regressions (primary)	0.3%	0.3%	1
Regressions (secondary)	0.2%	0.2%	1
Improvements (primary)	-0.3%	-0.4%	5
Improvements (secondary)	-0.4%	-0.9%	16
All (primary)	-0.2%	-0.4%	6

already triaged by nnethercote: “There’s a single regressing result and 25 improved results. The changes are all very small. This is fine”

Thin AssocItem #95884 (Comparison Link)

	mean	max	count
Regressions (primary)	0.3%	0.4%	16
Regressions (secondary)	0.5%	0.6%	9
Improvements (primary)	-0.3%	-0.4%	7
Improvements (secondary)	-0.4%	-0.5%	8
All (primary)	0.1%	0.4%	23

already triaged by nnethercote: “This looks perf-neutral, with the improvements roughly the same size as the regressions.”

Remove fn backtrace and replace with usages of provider API #99431 (Comparison Link)

	mean	max	count
Regressions (primary)	N/A	N/A	0
Regressions (secondary)	1.5%	1.5%	2
Improvements (primary)	-0.9%	-1.4%	2
Improvements (secondary)	N/A	N/A	0
All (primary)	-0.9%	-1.4%	2

the improvements are to building primary benchmark cargo opt in incr-patch: println and incr-full, by -1.38% and -0.42% respectively . The regression was to the externs incr-full in debug and opt profiles, both by 1.52%.
this is entirely acceptable (and wasn’t marked as a performance regression by the bot in the first place)

Nominated Issues

T-compiler

“compiler-builtins CI fails on powerpc64” rust#99853
- Seems that - for this Tier 2 target - tests are slowly degrading
- Opened a topic on Zulip to get more eyeballs
- cc @Nikita Popov

RFC

No nominated RFCs for T-compiler this time.

Next week’s WG checkins

@_WG-diagnostics by @Esteban Küber and @oli
@_WG-rustc-dev-guide by @Santiago Pastorino and @Yuki Okushi|217081