rustdoc-search: adjust packing heuristics to save more space

[notriddle]

This upgrade causes stringdex to use fewer, larger files with the suffix-only children included more often. This is particularly aimed at reducing the number of files required by many-assoc-items.

The cost of this change is that the browser has to download more data that it isn't using while walking the search tree, because branches of the search tree that it isn't using are in the files it downloads. When I tested against the wordnet dictionary, a search for "indexing" required 12MiB of transfer, where it used to require 8MiB.

#144476 (comment)

[notriddle]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: b7f830d (b7f830d29248bb15ffccc37d1d5ddb504b80808e, parent: 05f5a58e84a9c3a68586d70bf3d7442c571e379e)

[rust-timer]

Finished benchmarking commit (b7f830d): comparison URL.

Overall result: ❌ regressions - please read the text below

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	5.1%	[0.5%, 41.2%]	10
Regressions ❌ (secondary)	12.4%	[1.2%, 34.2%]	3
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	5.1%	[0.5%, 41.2%]	10

Max RSS (memory usage)

Results (primary -3.9%)

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)	-3.9%	[-3.9%, -3.9%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-3.9%	[-3.9%, -3.9%]	1

Cycles

Results (primary 18.2%, secondary 14.0%)

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

	mean	range	count
Regressions ❌ (primary)	18.2%	[18.2%, 18.2%]	1
Regressions ❌ (secondary)	14.0%	[14.0%, 14.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	18.2%	[18.2%, 18.2%]	1

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 472.158s -> 470.896s (-0.27%)
Artifact size: 378.09 MiB -> 378.09 MiB (-0.00%)

[notriddle]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: e946368 (e946368b895f934127080fbcdab45309179161e0, parent: d20509c2a0c71b60aa2b51566e4d14920e8a1661)

[rust-timer]

Finished benchmarking commit (e946368): comparison URL.

Overall result: ❌ regressions - please read the text below

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	0.5%	[0.4%, 0.7%]	6
Regressions ❌ (secondary)	2.5%	[1.2%, 3.8%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.5%	[0.4%, 0.7%]	6

Max RSS (memory usage)

Results (secondary -0.2%)

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)	5.1%	[5.1%, 5.1%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.9%	[-3.6%, -2.1%]	2
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary 2.7%)

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)	2.7%	[2.7%, 2.7%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 468.202s -> 466.275s (-0.41%)
Artifact size: 378.26 MiB -> 378.24 MiB (-0.01%)

[notriddle]

Okay, that seems like an acceptable tradeoff for the doc_files_count that this PR brings.

r? @GuillaumeGomez

[rustbot]

These commits modify the Cargo.lock file. Unintentional changes to Cargo.lock can be introduced when switching branches and rebasing PRs.

If this was unintentional then you should revert the changes before this PR is merged.
Otherwise, you can ignore this comment.

[GuillaumeGomez]

So it reduces the number of files on the disk but increases the network data usage? Not sure it's a good trace. :-/

Also, the previous PR already had up to 50% more instructions, so adding even more on top of it really doesn't seem that great.