2022-05-10 Triage Log

A good week: Several performance improvements, many around macro expansion. Only one regression of note, and that PR author opened an issue to follow up on it.

Triage done by @pnkfelix. Revision range: 468492c2..c51871c4

Summary:

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	11	37	117	65	128
mean	0.7%	0.7%	-1.2%	-1.6%	-1.1%
max	1.5%	1.9%	-6.5%	-5.2%	-6.5%

2 Regressions, 4 Improvements, 1 Mixed; 1 of them in rollups 59 artifact comparisons made in total

Regressions

Add a new Rust attribute to support embedding debugger visualizers #91779 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	88	52	0	0	88
mean	0.4%	0.5%	N/A	N/A	0.4%
max	0.9%	1.0%	N/A	N/A	0.9%

• Original PR author opened issue #96786 to investigate these regressions, and marked as triaged.
• (PR #96795 claims to resolve some of the regressions, but rust-timer didn't see any significant change from that.)

Rollup of 6 pull requests #96877 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	18	0	0	0	18
mean	0.4%	N/A	N/A	N/A	0.4%
max	0.7%	N/A	N/A	N/A	0.7%

• inspected graphs for the benchmarks at top of list of regressions: webrender-2022, regex-1.5.5, clap-3.1.6, cargo-0.60.0. It seems like any long-term effect here is quite minor, and the fact that this PR was tagged as significant is due to a blip in the measurements.
• marking as triaged.

Improvements

Speed up Token::{ident,lifetime} #96683 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	0	0	151	82	151
mean	N/A	N/A	-0.9%	-0.7%	-0.9%
max	N/A	N/A	-6.9%	-4.0%	-6.9%

Overhaul MacArgs #96546 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	1	1	16	9	17
mean	0.3%	0.2%	-0.8%	-0.2%	-0.8%
max	0.3%	0.2%	-1.4%	-0.3%	-1.4%

Remove ItemLikeVisitor impls from rustc_typeck #96531 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	0	0	9	9	9
mean	N/A	N/A	-0.3%	-0.4%	-0.3%
max	N/A	N/A	-0.4%	-0.6%	-0.4%

Optimize switch sources representation and usage #96838 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	0	0	4	7	4
mean	N/A	N/A	-0.7%	-2.0%	-0.7%
max	N/A	N/A	-1.0%	-2.6%	-1.0%

Mixed

store codegen_fn_attrs in crate metadata #96473 (Comparison Link)

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count	6	28	89	39	95
mean	0.3%	0.6%	-0.5%	-1.6%	-0.4%
max	0.6%	2.2%	-1.4%	-4.8%	-1.4%

• the regressions here were anticipated during development. The only regressions that arose were from doc builds (which I confirmed via inspection of the 6 regressing cases).