[borrowck] Do an in-place update of SCCs rather than recomputing

[amandasystems]

Exploit some of the nice properties of 'static in the constraint graph to avoid doing a full recomputation of SCC components when adding : 'static outlives.

Notably, this relies on the fact that adding any 'r: 'static will simply remove 'r's SCC as well as any of its predecessors from the set of SCCs, since they now all participate in a cycle with 'static (which outlives everything and thus has an edge into every region).

r? @lcnr

(who may or may not want to delegate this)

This almost entirely a performance thing so it wants a perf run, but it actually makes some of the code more straightforward.

[rustbot]

r? @jdonszelmann

rustbot has assigned @jdonszelmann.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 73 candidates
• Random selection from 21 candidates

[lcnr]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 6dbb4d3 (6dbb4d39316caf969b5bf65444b29a9e255a2d57, parent: 03c609abb6638f9d7f49f34326d4137d07f5cd61)

[rust-timer]

Finished benchmarking commit (6dbb4d3): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking means the PR may be perf-sensitive. Consider adding rollup=never if this change is not fit for rolling up.

@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This perf run didn't have relevant results for this metric.

Max RSS (memory usage)

Results (secondary -2.2%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.2%	[-2.2%, -2.2%]	1
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary 3.2%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.2%	[3.2%, 3.2%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 487.089s -> 490.144s (0.63%)
Artifact size: 393.44 MiB -> 393.41 MiB (-0.01%)

[amandasystems]

Wow. Ok, then I don't see any point in bothering and I can finally lay this intrusive idea to rest. Thanks for coming to my TED talk!

[amandasystems]

Probably what you really want is a specialised SCC construction for the outlives graph. That would allow tacking on all the extra annotations (and representatives etc) in a nicer way. It would also allow handling the extra :static constraints during construction. 'static can also be handled as a special case without having to add every SCC as a successor to it explicitly. However, I suspect the potential gains are at best very small for what is a large refactor.

I'm leaving this idea here for anyone who wants to investigate it. Ping me if you do, I'd like to see how it pans out!