[jdalton]

Validation update

Updated to current master. Local checks, 420 Node tests, 46 WPT pages, and 22 package tests pass. Coverage passes increased thresholds: statements 77.60%, branches 65.99%, functions 81.48%, lines 75.57%. Engine coverage comes from WPT; adapter coverage comes from Node.

New unit and WPT regressions cover generation promotion and eviction, capacity, updates, clearing, special keys, and the missing-Map fallback. They also fix a capacity bug: promotions now rotate a full young generation before the next insertion can exceed the limit.

Change

A strict LRU reorders on use and evicts one entry per insertion, both with Map delete, and V8 keeps a deleted entry until the map rehashes — so finding the oldest entry walks the tombstones every earlier eviction left, which profiled at 28% of an eviction-heavy run. Two generations replace it: the young one fills, becomes the old one, and the previous old one is dropped whole, so eviction is a pointer swap.

Detail, and how it was checked

A strict LRU reorders on use and evicts one entry per insertion, both with Map.delete, and V8 keeps a deleted entry in the backing store until the map rehashes — so keys().next(), the way the oldest entry is found, walks the tombstones every earlier eviction left. Profiling 8000 selectors cycling through a 4096-entry cache put Map.set at 28% of total run time.

Entries are now written to a young generation. When it fills, it becomes the old generation and the previous old one is dropped whole: no per-insertion delete, no iteration, and eviction is a pointer swap. A hit in the old generation carries the entry back, so anything still in use survives the next swap. Capacity is unchanged, half the limit per generation.

get() also stops calling has() first. A cached value is never undefined, so one lookup answers both whether the entry exists and what it holds.

Historical measurements from the original proposal, before the current integration and capacity fix. These have not been rerun for this revision:

workload	master	with this patch
30 selectors, all hit	4.68 us	3.13 us	1.50x
2000 selectors	690 us	732 us	0.94x
3000 selectors	22.72 ms	973 us	23.34x
8000 selectors	67.60 ms	34.86 ms	1.94x

The loss is a working set that straddles a generation: it no longer fits the young one, so a pass takes old-generation hits and pays to carry them across. At 3000 the comparison inverts, because a selector is not one cache entry — a :not() argument takes its own — so 3000 selectors overflow a 4096-entry LRU while the segmented cache degrades instead of thrashing.

Extracted from #167 as a standalone cache change.

References: the spec, the browser source, and what each part was reasoned from

• MDN: Map — the delete semantics this replaces.
• Reading: zod.dev/blog/reducing-memory-footprint — the same measurement discipline applied to a library that caches heavily.

Minimums are now statements 77%, branches 65%, functions 81%, and lines 75%. Added WPT regressions for query-plan eviction across all four caches, callbacks after mutation, clearing, missing WeakMap, and single-entry updates.

[jdalton]

This one sits with #186 and #188, which touch the resolver caches.

They do not depend on each other. All seventeen in the series cherry-pick onto master in any order, and I checked that in both directions, so any one of these can land alone. The order below is the one they read best in:

• #185 makes eviction cheap, by replacing the LRU that walked tombstones with two generations.
• #186 then raises the limit to 4096, which is worth doing once eviction no longer costs per insertion.
• #188 caches the query plan rather than the answer, so a cache entry stops retaining DOM.

[jdalton]

Fixed in f22b2fa. Updated to master and moved the legacy cache tests into the active test layout. New unit and WPT regressions cover cache misses, updates, promotion, eviction, clearing, special keys, and the missing-Map fallback. They exposed and now protect a capacity bug: promoting old entries could let the next insertion exceed 1,000 entries. Promotions now rotate a full young generation.

Local checks and all 22 package tests pass. All 417 Node tests and 45 WPT pages pass. Coverage passes unchanged thresholds: statements 77.22%, branches 64.87%, functions 79.85%, lines 75.20%. Engine coverage still comes exclusively from the WPT runner. CI is running.