docs: fix factual errors found in review

[lesnik512]

Summary

Fixes docs claims that disagree with the code. Each item below names what was wrong and the source that shows it.

Changes

• troubleshooting/unsupported-creator-parameter-error.md, providers/factories.md: the example creator was positional-only, but for def f(items, /) the kwargs route still raises UnsupportedCreatorParameterError and skip_creator_parsing raises CreatorCallError (types_parser.py:65-76, confirmed by running). Dropped the / from the example and stated that positional-only params need a default.
• introduction/design-decisions.md: a Factory subclass raises TypeError (resolver_compiler.py:48-57), so the generator-creator advice now points to a plain creator plus CacheSettings(finalizer=). A missing provider is an ArgumentResolutionError from validate() or resolve (factory.py:190), not a declaration-time error. Overrides use one tree-wide registry and the lock is per tree (container.py:116-126). The async and Beta statements are now plain.
• migration/from-that-depends.md, from-dependency-injector.md: a cached Factory caches one object per container, so the ThreadLocalSingleton workaround now uses an uncached Factory over a module-level threading.local() (tested with 3 threads).
• migration/to-3.x.md: added the blank line that ends the !!! warning block, and removed the "an earlier version said" note.
• troubleshooting/max-scope-reached-error.md: uses build_child_container(scope=MyScope.SUBSTEP) with a one-member IntEnum in place of the internal parent_container= (confirmed by running).
• troubleshooting/circular-dependency.md, missing-provider.md, context-not-set.md: sample output regenerated from real runs, now with the scope column and (module:line) (exceptions/rendering.py:42).
• index.md: Flask, gRPC and Typer don't close the root container (no shutdown hook), and pytest builds no child per test. Also "smallest example".
• dev/contributing.md: the gate is report.fail_under = 100 in pyproject.toml. providers/errors-and-exceptions.md: Group(scope=...) also raises InvalidScopeTypeError (group.py:23). recipes/good-and-bad-practices.md: removed the history wording.
• Integrations, checked against each modern-di-<name> repo:
  • celery: worker_init/worker_shutdown (main.py:37-40), plus the *args/**kwargs TypeError (main.py:68-75).
  • arq: for @inject tasks the wrapper closes the child, and on_job_end only closes it as a safety net (main.py:86-90,188-198).
  • flask: corrected the "as above" reference and the route-order comment (route order only matters for auto_inject, main.py:51-52).
  • grpc: validate() never constructs anything, so the default only matters for resolving outside an RPC (confirmed by running).
  • fastapi: added fetch_di_container to the API table. typer: added action_scope(ctx).
  • faststream: corrected the FromDI signature and dropped "no longer". taskiq: added use_cache, and the child is built lazily via TaskiqDepends.
  • litestar: name collisions with app dependencies, and passing a provider instance to FromDI under autowired_groups raises ImproperlyConfiguredException.
  • pytest: an override set in the test body landed after the fixture had resolved, so the example failed when run. Moved the override into a container fixture that the dependency fixture uses (confirmed with pytest).
  • writing-integrations: the pin is now <4, Typer root lookup goes through context_settings["obj"], decorator integrations do register connection providers, and noted auto_inject/DITask and the extra modules in aiogram and pytest.
• Checked and left unchanged: providers/scopes.md:16. The FastAPI, Starlette, Litestar and aiohttp integrations do enter SESSION for websockets.

Checklist

• Lint and format pass (ruff)
• Type check passes (ty)
• Tests pass and new behavior is covered
• Build succeeds (uv build) if packaging or build config changed
• Repo metadata stays consistent across the three surfaces (GitHub description, pyproject description, profile blurb) if this touches packaging

just lint-ci, just test-ci (100% coverage) and mkdocs build --strict all pass.

[github-actions]

Benchmark

Details

Benchmark suite	Current: 8a17acb	Previous: 8ce868f	Ratio
benchmarks/test_guard_by_type.py::test_g16_resolve_by_type	6860220.469637784 iter/sec (stddev: 1.1350279026376358e-8)	3442054.526547969 iter/sec (stddev: 2.143375745788722e-8)	0.50
benchmarks/test_guard_by_type.py::test_g17_resolve_by_type_large_registry	6634242.910004253 iter/sec (stddev: 1.1118838523911411e-8)	3548434.840392541 iter/sec (stddev: 1.2827232048867815e-8)	0.53
benchmarks/test_guard_cold.py::test_g8_cold_first_resolve	20086.76388833524 iter/sec (stddev: 0.0002127862113507429)	15518.545714422256 iter/sec (stddev: 0.00004698793690022765)	0.77
benchmarks/test_guard_cold.py::test_g8b_cold_first_resolve_cached	18162.24197549022 iter/sec (stddev: 0.000032183694633676134)	13185.693699108479 iter/sec (stddev: 0.00010949077679688568)	0.73
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[1]	802.4842572229129 iter/sec (stddev: 0.00009788610786291962)	404.3230311283974 iter/sec (stddev: 0.00007048845657185601)	0.50
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[2]	682.3258283587447 iter/sec (stddev: 0.00004926030062943845)	383.4654873688378 iter/sec (stddev: 0.000046998100263502144)	0.56
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[4]	603.8768218237799 iter/sec (stddev: 0.00006520483457299476)	352.2463486495161 iter/sec (stddev: 0.00009127243270054909)	0.58
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[1]	1992.770097231849 iter/sec (stddev: 0.0001333977483374318)	1626.7073683300935 iter/sec (stddev: 0.0001986442227433345)	0.82
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[2]	1625.6971565124065 iter/sec (stddev: 0.00013948311717709868)	1327.7632191013151 iter/sec (stddev: 0.00017288226483966492)	0.82
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[4]	970.6024731749375 iter/sec (stddev: 0.001315486110555091)	960.0010675205933 iter/sec (stddev: 0.00017975799786761426)	0.99
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[1]	2163.313434698019 iter/sec (stddev: 0.00013049710007195453)	1407.057453110696 iter/sec (stddev: 0.0010831166068909466)	0.65
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[2]	1439.2394483119317 iter/sec (stddev: 0.00015047467334073196)	1175.7928128429965 iter/sec (stddev: 0.00021455490282242838)	0.82
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[4]	957.3024442464645 iter/sec (stddev: 0.0001566678311087579)	784.9046406926733 iter/sec (stddev: 0.00021119103540343572)	0.82
benchmarks/test_guard_lifecycle.py::test_g6_build_child_container	1268532.4262865619 iter/sec (stddev: 4.743103233469329e-8)	826944.4808698264 iter/sec (stddev: 4.169575359621365e-8)	0.65
benchmarks/test_guard_lifecycle.py::test_g6b_build_child_container_auto_scope	1149102.8781506433 iter/sec (stddev: 3.822966211174696e-8)	754895.2503574233 iter/sec (stddev: 3.0409338412035246e-8)	0.66
benchmarks/test_guard_lifecycle.py::test_g7_request_lifecycle_batch	3171.698285863586 iter/sec (stddev: 0.000007740096462751993)	2421.0815300750833 iter/sec (stddev: 0.000020424490014428993)	0.76
benchmarks/test_guard_lifecycle.py::test_g7c_event_loop_floor_control	93277.60687478787 iter/sec (stddev: 0.00000100094982014164)	58408.76624586848 iter/sec (stddev: 0.0000018447825779836061)	0.63
benchmarks/test_guard_lifecycle.py::test_g13_teardown_at_scale	63873.447433154484 iter/sec (stddev: 0.0000013026365585484084)	43876.32191693938 iter/sec (stddev: 0.0000021012125198676007)	0.69
benchmarks/test_guard_lifecycle.py::test_g13b_teardown_at_scale_async_no_finalizers	806.807827234481 iter/sec (stddev: 0.00004342058717888897)	553.8315819315635 iter/sec (stddev: 0.00011049984251667817)	0.69
benchmarks/test_guard_resolve.py::test_g1_transient_resolve	3751181.5049841 iter/sec (stddev: 3.642543373787177e-8)	2179414.0503717246 iter/sec (stddev: 3.2116689283894473e-8)	0.58
benchmarks/test_guard_resolve.py::test_g2_cached_resolve	6474185.673742138 iter/sec (stddev: 1.1727614617152446e-8)	3331944.800863088 iter/sec (stddev: 1.8996420430089775e-8)	0.51
benchmarks/test_guard_resolve.py::test_g3_deep_chain	1270769.8597294868 iter/sec (stddev: 4.8332149085018514e-8)	728489.2646086639 iter/sec (stddev: 6.251243141025456e-8)	0.57
benchmarks/test_guard_resolve.py::test_g4_wide_resolve	764213.5894289435 iter/sec (stddev: 2.5539878049722226e-7)	468680.21058674285 iter/sec (stddev: 3.2602527219035675e-7)	0.61
benchmarks/test_guard_resolve.py::test_g5_cross_scope	3550509.7289320664 iter/sec (stddev: 2.3962636628665118e-8)	1866879.1921924949 iter/sec (stddev: 3.339221547534178e-8)	0.53
benchmarks/test_guard_resolve.py::test_g9_context_resolve	1768069.582736624 iter/sec (stddev: 1.2523246341256786e-7)	1052288.8650783815 iter/sec (stddev: 1.333290563243788e-7)	0.60
benchmarks/test_guard_resolve.py::test_g12_override_active_resolve	1296090.013971847 iter/sec (stddev: 3.4713002666392195e-8)	720244.7622185523 iter/sec (stddev: 6.1118782837122e-8)	0.56
benchmarks/test_guard_resolve.py::test_g18_alias_hop	4519560.54503975 iter/sec (stddev: 1.5636115066043727e-8)	1889630.0710914063 iter/sec (stddev: 4.658060565489605e-8)	0.42
benchmarks/test_guard_validate.py::test_g10_validate_deep_chain	28760.641820904686 iter/sec (stddev: 0.0000154937778851691)	22333.268360870607 iter/sec (stddev: 0.00003390432019260961)	0.78
benchmarks/test_guard_validate.py::test_g11_validate_wide	18381.735260884343 iter/sec (stddev: 0.000018514738717146934)	14119.363707672732 iter/sec (stddev: 0.000028698284721104412)	0.77

This comment was automatically generated by workflow using github-action-benchmark.