test(pm): cover resolver mainloop and remove preload

[elrrrrrrr]

Summary

• Add demand mainloop unit coverage.
• Remove obsolete resolver preload events/module.

Review Focus

• Mainloop invariants and preload API removal impact.

[gemini-code-assist]

Code Review

This pull request removes the preload module and its associated build events, streamlining the dependency resolution process. It introduces significant test coverage in builder.rs for manifest state handling, fetch queue prioritization, and concurrency limits. The reviewer noted that the CountingRegistry mock implementation in the tests is missing the #[async_trait::async_trait] attribute on the RegistryClient trait implementation, which is necessary for consistency with other async traits in the module.

[gemini-code-assist]

The implementation of RegistryClient for CountingRegistry is missing the #[async_trait::async_trait] attribute. Since the ManifestProvider implementation for the same struct (line 1954) uses it, and the methods are async, this attribute is likely required for the trait to be correctly implemented and to maintain consistency with other async trait implementations in this module.

    #[async_trait::async_trait]
    impl crate::traits::registry::RegistryClient for CountingRegistry {

[github-actions]

📊 pm-bench-phases · afb9163 · linux (ubuntu-latest)

Workflow run — ant-design

PMs: utoo (this branch) · utoo-npm (latest published) · bun (latest)

npmjs.org

p0_full_cold

PM	wall	±σ	user	sys	RSS	pgMinor
bun	9.37s	0.18s	10.54s	10.51s	591M	303.2K
utoo-next	8.78s	0.62s	10.61s	12.61s	977M	129.6K
utoo-npm	8.47s	0.06s	10.78s	12.71s	981M	130.5K
utoo	8.39s	0.19s	11.46s	12.69s	1003M	158.2K

PM	vCtx	iCtx	netRX	netTX	cache	node_mod	lock
bun	19.0K	18.0K	1.22G	7M	1.93G	1.81G	1M
utoo-next	122.8K	88.0K	1.19G	5M	1.77G	1.77G	2M
utoo-npm	129.4K	95.1K	1.19G	5M	1.77G	1.77G	2M
utoo	111.5K	71.5K	1.19G	6M	1.77G	1.76G	2M

p1_resolve

PM	wall	±σ	user	sys	RSS	pgMinor
bun	2.54s	0.06s	3.98s	1.16s	489M	192.1K
utoo-next	3.29s	0.08s	5.48s	1.86s	613M	84.0K
utoo-npm	3.43s	0.01s	5.71s	2.25s	619M	86.0K
utoo	2.64s	0.07s	6.04s	1.80s	657M	123.4K

PM	vCtx	iCtx	netRX	netTX	cache	node_mod	lock
bun	13.7K	3.4K	204M	3M	109M	-	1M
utoo-next	55.7K	75.0K	202M	3M	7M	3M	2M
utoo-npm	81.9K	101.0K	203M	3M	7M	3M	2M
utoo	17.4K	21.4K	205M	3M	7M	3M	2M

p3_cold_install

PM	wall	±σ	user	sys	RSS	pgMinor
bun	6.92s	0.14s	6.56s	10.42s	572M	208.1K
utoo-next	7.18s	1.29s	5.16s	11.09s	497M	65.0K
utoo-npm	6.52s	0.27s	5.17s	10.98s	489M	64.7K
utoo	7.03s	0.93s	5.14s	11.18s	495M	66.1K

PM	vCtx	iCtx	netRX	netTX	cache	node_mod	lock
bun	8.5K	8.7K	1.02G	5M	1.82G	1.82G	1M
utoo-next	103.8K	49.9K	1017M	3M	1.76G	1.76G	3M
utoo-npm	97.6K	49.2K	1017M	2M	1.76G	1.76G	3M
utoo	111.0K	54.0K	1017M	3M	1.76G	1.76G	3M

p4_warm_link

PM	wall	±σ	user	sys	RSS	pgMinor
bun	3.38s	0.09s	0.23s	2.36s	136M	32.3K
utoo-next	2.58s	0.41s	0.50s	3.90s	81M	18.9K
utoo-npm	2.34s	0.03s	0.51s	3.89s	81M	18.8K
utoo	2.24s	0.06s	0.51s	3.87s	81M	18.6K

PM	vCtx	iCtx	netRX	netTX	cache	node_mod	lock
bun	427	22	5M	31K	1.98G	1.81G	1M
utoo-next	42.0K	17.8K	307K	12K	1.76G	1.76G	2M
utoo-npm	43.9K	20.6K	307K	28K	1.76G	1.76G	2M
utoo	42.7K	19.4K	309K	28K	1.77G	1.76G	2M

npmmirror.com: no output captured.

[elrrrrrrr]

Superseded: re-splitting #3028 from scratch (mirroring the #3029 installer split into #3075/#3076). Closing this draft to avoid a stale parallel split stack.