Adopt synchronous machine transitions

[SandroMaglione]

Summary

• replace Effectful transition, lifecycle, initial, choice, and history callbacks with a synchronous transition core
• add a closed, protocol-typed enqueue API for raise, emit, sendTo, and stop
• keep asynchronous work and Effect service/error inference at invoke, actor, and child-machine boundaries
• migrate runtime execution, Cluster support, testing utilities, examples, documentation, consumer fixtures, and type-performance fixtures
• add differential coverage for planning/runtime/resume ordering and large raised-event drainage
• add memory-attribution profiles for Effect primitives, managed processes, statecharts, parent-child ownership, and observation

Changeset

• Added or updated for a library or package-metadata change
• Not required because this PR does not change src/ or package.json

Validation

• pnpm check
• Relevant example checks, when examples changed
• Reviewed the automated type-performance report, when the public TypeScript API or inference changed
• Reviewed the automated runtime-performance report, when runtime behavior changed

Local runtime comparison used three paired full runs on Apple M1 with Node 26.5.0. Median gains over the exact parent commit were 3.42x for pure planning, 2.25x for runtime bursts, 2.07x for observed bursts, and 2.30x for child sends. Idle memory was effectively unchanged.

The full memory ladder measured 0.7 KiB per suspended Effect fiber, 2.8 KiB per Effect queue/waiting-fiber mailbox, 3.2 KiB per minimal Effect actor shell, 18.6 KiB per raw managed process, 23.5 KiB per idle statechart, 46.1 KiB per two independent statecharts, 74.1 KiB per parent-child pair, and 82.9 KiB with child observation active. These attribution profiles are now part of the local and pull-request benchmark reports.

[github-actions]

Type performance

Measured with TypeScript 6.0.3 and skipLibCheck=true.

Scenario	Base	PR	Difference
Effect only	55	55	0 (0.0%)
Import effect-machine	55	55	0 (0.0%)
Machine.defineStates (3 states)	2,965	2,965	0 (0.0%)
Machine.make (3 states, 2 events)	8,731	8,715	-16 (-0.2%)
machine.handle (3 states, 2 transitions)	23,634	23,492	-142 (-0.6%)
machine.handle (depth 8)	113,023	117,075	+4,052 (+3.6%)
machine.handle (depth 12)	125,661	129,713	+4,052 (+3.2%)
machine.handle (depth 16)	141,003	145,055	+4,052 (+2.9%)
machine.handle (depth 24)	179,799	183,851	+4,052 (+2.3%)
machine.handle (wide depth 16)	210,137	214,189	+4,052 (+1.9%)
machine.handle (parallel/history/choice)	128,880	132,667	+3,787 (+2.9%)
machine.handle (4 successive calls)	129,498	128,674	-824 (-0.6%)
machine exact input/output/error/services	107,375	110,584	+3,209 (+3.0%)
execution adapter readiness	122,890	127,854	+4,964 (+4.0%)

Marginal instantiations are measured against the matching setup without that API call:

Scenario	Base	PR	Difference
Import effect-machine	0	0	0
Machine.defineStates (3 states)	2,910	2,910	0 (0.0%)
Machine.make (3 states, 2 events)	5,758	5,742	-16 (-0.3%)
machine.handle (3 states, 2 transitions)	14,903	14,777	-126 (-0.8%)
machine.handle (depth 8)	105,546	109,601	+4,055 (+3.8%)
machine.handle (depth 12)	116,984	121,039	+4,055 (+3.5%)
machine.handle (depth 16)	131,126	135,181	+4,055 (+3.1%)
machine.handle (depth 24)	167,522	171,577	+4,055 (+2.4%)
machine.handle (wide depth 16)	198,169	202,224	+4,055 (+2.0%)
machine.handle (parallel/history/choice)	110,514	114,331	+3,817 (+3.5%)
machine.handle (4 successive calls)	114,497	113,703	-794 (-0.7%)
machine exact input/output/error/services	95,819	100,670	+4,851 (+5.1%)
execution adapter readiness	96,968	102,092	+5,124 (+5.3%)

Check times (informational)

Scenario	Base	PR
Effect only	0.03s	0.03s
Import effect-machine	0.03s	0.03s
Machine.defineStates (3 states)	0.10s	0.10s
Machine.make (3 states, 2 events)	0.16s	0.16s
machine.handle (3 states, 2 transitions)	0.24s	0.24s
machine.handle (depth 8)	0.53s	0.51s
machine.handle (depth 12)	0.56s	0.56s
machine.handle (depth 16)	0.62s	0.60s
machine.handle (depth 24)	0.70s	0.68s
machine.handle (wide depth 16)	0.76s	0.77s
machine.handle (parallel/history/choice)	0.63s	0.63s
machine.handle (4 successive calls)	0.60s	0.59s
machine exact input/output/error/services	0.55s	0.52s
execution adapter readiness	0.56s	0.59s

Type instantiations are the comparison metric. Check time varies with runner load and is informational only.

[github-actions]

Runtime performance

Median of 5 independent benchmark processes on AMD EPYC 9V74 80-Core Processor with Node v24.18.0.

Pull request baseline

Scenario	Effect Machine	XState 5	XState 6 alpha
Plan counter transitions	108,366 transitions/s	138,975 transitions/s	115,785 transitions/s
Drain burst with terminal fence	65,283 increments/s	388,989 increments/s	107,642 increments/s
Drain burst with a change observer	55,099 increments/s	384,952 increments/s	107,375 increments/s
Lookup and send to one child	58,366 increments/s	384,667 increments/s	104,826 increments/s
Start and stop a machine	15,240 machines/s	179,565 machines/s	73,692 machines/s
Start and stop a parent with one child	4,679 families/s	71,526 families/s	30,574 families/s

Memory profile	Effect Machine	XState 5	XState 6 alpha
Idle machine	23.4 KiB	3.8 KiB	6.1 KiB
Raw managed process	18.4 KiB	—	—
Two independent idle machines	46.0 KiB	7.3 KiB	11.8 KiB
Idle parent with one child	73.7 KiB	5.5 KiB	8.8 KiB
Parent with observed child registry	82.6 KiB	—	—

Effect Machine change from base

Metric	Base	Base variability	PR	PR variability	Difference
Plan counter transitions	38,977 transitions/s	0.5% MAD	108,366 transitions/s	0.4% MAD	+178.0%
Drain burst with terminal fence	33,562 increments/s	0.6% MAD	65,283 increments/s	0.9% MAD	+94.5%
Drain burst with a change observer	30,603 increments/s	0.4% MAD	55,099 increments/s	1.8% MAD	+80.0%
Lookup and send to one child	28,837 increments/s	0.3% MAD	58,366 increments/s	0.5% MAD	+102.4%
Start and stop a machine	10,186 machines/s	1.9% MAD	15,240 machines/s	1.4% MAD	+49.6%
Start and stop a parent with one child	3,920 families/s	1.6% MAD	4,679 families/s	1.0% MAD	+19.4%
Idle machine heap per unit	23.6 KiB	0.0% MAD	23.4 KiB	0.0% MAD	-1.1%
Raw managed process heap per unit	18.4 KiB	0.0% MAD	18.4 KiB	0.0% MAD	-0.1%
Two independent idle machines heap per unit	46.3 KiB	0.0% MAD	46.0 KiB	0.0% MAD	-0.7%
Idle parent with one child heap per unit	73.9 KiB	0.0% MAD	73.7 KiB	0.0% MAD	-0.2%
Parent with observed child registry heap per unit	82.8 KiB	0.0% MAD	82.6 KiB	0.0% MAD	-0.2%

Versions and interpretation

• Effect Machine: 0.3.0
• XState 5: 5.32.5
• XState 6 alpha: 6.0.0-alpha.27

Higher throughput is better; lower heap is better. Variability is the median absolute deviation across independent processes, relative to their median. Runtime measurements on shared GitHub-hosted hardware remain informational, so small differences should be confirmed across multiple workflow runs.