fix: dispose factory outputs hidden behind a non-disposable return type

[vbreuss]

A factory registration computes disposability from the producer method's declared return type, so a factory declared static IFoo Create() that builds a DisposableFooImpl : IFoo, IDisposable was flagged non-disposable and never tracked, leaking the instance when the owning scope or root tore down. Constructed types use the concrete info.Symbol (always truthful) and a pre-built Instance is intentionally not owned, so the lie is possible only for factory production.

For a factory whose declared return type is not itself IDisposable yet could produce one at runtime (an interface, a type parameter, or a non-sealed class), the emitted resolver now tracks the realized instance behind a runtime is IDisposable check instead of trusting the static flag. This retains only genuinely-disposable outputs (no retention cost otherwise) and closes the gap across the synchronous fresh resolver, the async fresh/caching registration, and the caching singleton/scoped disposal add. A sealed return type that is not IDisposable cannot hide one, so it emits no check (and a runtime is IDisposable against it would not even compile). The existing __gate lock and __disposed re-check semantics are preserved for factory outputs, so one built during a concurrent dispose is still disposed rather than leaked.

The static InstanceModel.IsDisposable semantics are unchanged: the AWT118 root-accumulation analyzer and strict-lifetime withholding keep their compile-time behavior. AWT118 still under-fires for factory-hidden disposables (the same declared-type gap, but compile-time) - that is left to a separate body-analysis linter PR. IAsyncDisposable remains out of scope, matching the current IDisposable-only drain.

[github-actions]

Test Results

   18 files  ± 0     18 suites  ±0   1m 1s ⏱️ +2s
  213 tests + 8    212 ✅ + 8  1 💤 ±0  0 ❌ ±0 
1 091 runs  +48  1 090 ✅ +48  1 💤 ±0  0 ❌ ±0 

Results for commit 959470b. ± Comparison against base commit 448c1d0.

♻️ This comment has been updated with latest results.

[github-actions]

🚀 Benchmark Results

Details

BenchmarkDotNet v0.15.8, Linux Ubuntu 24.04.4 LTS (Noble Numbat)

Intel Xeon Platinum 8370C CPU 2.80GHz (Max: 2.79GHz), 1 CPU, 4 logical and 2 physical cores

.NET SDK 10.0.301

[Host] : .NET 10.0.9 (10.0.9, 10.0.926.27113), X64 RyuJIT x86-64-v4
Job=InProcess  Toolchain=InProcessEmitToolchain  IterationCount=15

LaunchCount=1  WarmupCount=10

Resolve	Size	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
baseline*	8	7.021 ns	0.1950 ns	0.1728 ns	0.61	-	NA
Awaiten	8	11.522 ns	0.0380 ns	0.0336 ns	1.00	-	NA
MsDI	8	7.509 ns	0.0156 ns	0.0122 ns	0.65	-	NA
Autofac	8	141.659 ns	3.2337 ns	3.0248 ns	12.29	656 B	NA
Jab	8	3.438 ns	0.0081 ns	0.0072 ns	0.30	-	NA
PureDI	8	4.212 ns	0.0131 ns	0.0123 ns	0.37	-	NA
DryIoc	8	7.198 ns	0.0096 ns	0.0075 ns	0.62	-	NA
SimpleInjector	8	9.829 ns	0.0262 ns	0.0204 ns	0.85	-	NA
baseline*	256	168.130 ns	3.5820 ns	3.3506 ns	0.78	-	NA
Awaiten	256	215.755 ns	0.2778 ns	0.2320 ns	1.00	-	NA
MsDI	256	7.535 ns	0.0102 ns	0.0090 ns	0.03	-	NA
Autofac	256	141.756 ns	1.9778 ns	1.8501 ns	0.66	656 B	NA
Jab	256	77.509 ns	0.0698 ns	0.0653 ns	0.36	-	NA
PureDI	256	8.271 ns	0.0198 ns	0.0185 ns	0.04	-	NA
DryIoc	256	7.258 ns	0.0119 ns	0.0106 ns	0.03	-	NA
SimpleInjector	256	13.435 ns	0.0747 ns	0.0698 ns	0.06	-	NA

Details

BenchmarkDotNet v0.15.8, Linux Ubuntu 24.04.4 LTS (Noble Numbat)

Intel Xeon Platinum 8370C CPU 2.80GHz, 1 CPU, 4 logical and 2 physical cores

.NET SDK 10.0.301

[Host] : .NET 10.0.9 (10.0.9, 10.0.926.27113), X64 RyuJIT x86-64-v4
Job=InProcess  Toolchain=InProcessEmitToolchain  IterationCount=15

LaunchCount=1  WarmupCount=10

Realistic	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
baseline*	257.5 ns	1.06 ns	0.99 ns	0.91	560 B	1.00
Awaiten	281.9 ns	1.33 ns	1.24 ns	1.00	560 B	1.00
MsDI	701.9 ns	3.23 ns	2.86 ns	2.49	1104 B	1.97
Autofac	7,651.0 ns	15.82 ns	14.02 ns	27.14	13696 B	24.46
Jab	224.1 ns	0.38 ns	0.35 ns	0.79	432 B	0.77
DryIoc	438.0 ns	2.05 ns	1.92 ns	1.55	944 B	1.69
SimpleInjector	773.0 ns	1.67 ns	1.56 ns	2.74	1096 B	1.96
PureDI	199.9 ns	1.26 ns	1.18 ns	0.71	632 B	1.13

Details

BenchmarkDotNet v0.15.8, Linux Ubuntu 24.04.4 LTS (Noble Numbat)

AMD EPYC 7763 2.45GHz, 1 CPU, 4 logical and 2 physical cores

.NET SDK 10.0.301

[Host] : .NET 10.0.9 (10.0.9, 10.0.926.27113), X64 RyuJIT x86-64-v3
Job=InProcess  Toolchain=InProcessEmitToolchain  IterationCount=15

LaunchCount=1  WarmupCount=10

Build	Size	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
baseline*	8	17.658 ns	0.5781 ns	0.5408 ns	1.07	136 B	1.00
Awaiten	8	16.49 ns	0.345 ns	0.323 ns	1.00	136 B	1.00
MsDI	8	1,517.02 ns	25.035 ns	23.418 ns	92.04	5688 B	41.82
Autofac	8	28,772.08 ns	183.958 ns	172.075 ns	1,745.66	33098 B	243.37
Jab	8	10.76 ns	0.620 ns	0.580 ns	0.65	96 B	0.71
PureDI	8	15.04 ns	0.205 ns	0.181 ns	0.91	128 B	0.94
DryIoc	8	702.49 ns	12.964 ns	12.127 ns	42.62	1472 B	10.82
SimpleInjector	8	11,310.45 ns	58.279 ns	51.663 ns	686.23	24760 B	182.06
baseline*	256	79.822 ns	0.7048 ns	0.6248 ns	0.93	2120 B	1.00
Awaiten	256	85.64 ns	0.478 ns	0.447 ns	1.00	2120 B	1.00
MsDI	256	13,919.90 ns	34.823 ns	32.573 ns	162.53	61016 B	28.78
Autofac	256	678,016.30 ns	1,509.390 ns	1,260.409 ns	7,916.82	739637 B	348.89
Jab	256	70.36 ns	0.516 ns	0.483 ns	0.82	2080 B	0.98
PureDI	256	78.71 ns	1.601 ns	1.420 ns	0.92	2112 B	1.00
DryIoc	256	43,537.70 ns	840.858 ns	745.399 ns	508.37	80128 B	37.80
SimpleInjector	256	338,098.84 ns	2,824.230 ns	2,358.360 ns	3,947.79	573171 B	270.36

baseline* rows show the corresponding Awaiten benchmark from the most recent successful main branch build with results, for regression comparison.

[github-actions]

👽 Mutation Results

Awaiten

Details

File	Score	Killed	Survived	Timeout	No Coverage	Ignored	Compile Errors	Runtime Errors	Total Detected	Total Undetected	Total Mutants

The final mutation score is NaN%

Coverage Thresholds: high:80 low:60 break:0

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
87.5% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud