refactor(factory): name the parsed creator and the cache coercion

[lesnik512]

Closes #537.

Two names for two things Factory.__init__ carried by position or by ladder.

ParsedCreator

parse_creator returns a frozen, slotted dataclass with return_type, params and has_positional_only_gap. The docstring paragraph that explained the third tuple element now sits on the class. Factory.__init__ reads the fields by name, and the skip_creator_parsing branch builds an empty ParsedCreator so the assignments after the branch are one path instead of two sets of locals. The tests stop unpacking _ret, params, _gap.

parse_creator is not exported from modern_di, so its return type is not public API.

CacheSettings.coerce

@staticmethod
def coerce(cache: "bool | CacheSettings[types.T] | None") -> "CacheSettings[types.T] | None":
    if cache is True:
        return CacheSettings()
    return cache or None

A staticmethod over the invariant types.T rather than a classmethod over the class's covariant T_co: ty 0.0.84 rejects a covariant variable in a parameter position, which the first CI run caught.

Replaces the three-branch ladder and its two comments at the top of Factory.__init__. Two new tests pin the mapping (True to defaults, False and None to None, an instance returned as itself).

Behaviour

None at resolve time. One extra object per Factory at construction, which is import time for a Group body.

Checks

• Full suite with coverage: 548 passed, 8 skipped, 100% line coverage.
• ruff format, ruff check, eof-fixer and ty check clean with the locked tool versions.

[github-actions]

Benchmark

Details

Benchmark suite	Current: abc6954	Previous: 0a78def	Ratio
benchmarks/test_guard_by_type.py::test_g16_resolve_by_type	3471998.7122578854 iter/sec (stddev: 1.7036046574653683e-8)	4968591.5454160115 iter/sec (stddev: 8.41396506374104e-9)	1.43
benchmarks/test_guard_by_type.py::test_g17_resolve_by_type_large_registry	3487433.608401099 iter/sec (stddev: 1.1046607004951574e-8)	4996027.408800345 iter/sec (stddev: 1.4695694448414382e-8)	1.43
benchmarks/test_guard_cold.py::test_g8_cold_first_resolve	14624.276884190329 iter/sec (stddev: 0.000053495968236821545)	22539.3628633738 iter/sec (stddev: 0.00004662239001453159)	1.54
benchmarks/test_guard_cold.py::test_g8b_cold_first_resolve_cached	12739.26178959642 iter/sec (stddev: 0.00012195389633674608)	19191.6790496122 iter/sec (stddev: 0.0001426656267049193)	1.51
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[1]	374.1639969967372 iter/sec (stddev: 0.00004720803635724062)	630.6994413228471 iter/sec (stddev: 0.000033762402906412355)	1.69
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[2]	354.37649069828245 iter/sec (stddev: 0.000040725568567036396)	588.7689052241964 iter/sec (stddev: 0.0001210559570636086)	1.66
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[4]	329.0192872749762 iter/sec (stddev: 0.00004688377151189849)	537.4440834218 iter/sec (stddev: 0.000023129769302699955)	1.63
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[1]	1615.6028242953948 iter/sec (stddev: 0.00019030577561535814)	2310.1930027653325 iter/sec (stddev: 0.00013228844954993188)	1.43
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[2]	1278.324365602773 iter/sec (stddev: 0.00020869889101231932)	1813.3266361134993 iter/sec (stddev: 0.00015685191293119628)	1.42
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[4]	934.2591280741457 iter/sec (stddev: 0.00021793865330889902)	1259.7266512835552 iter/sec (stddev: 0.00019618441960171664)	1.35
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[1]	1359.737131992806 iter/sec (stddev: 0.0011082844760762311)	1659.5303634170655 iter/sec (stddev: 0.0016704704107851974)	1.22
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[2]	1169.5847953512155 iter/sec (stddev: 0.0002133960236955586)	1606.1059651690402 iter/sec (stddev: 0.00018955869305135043)	1.37
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[4]	744.5961723917271 iter/sec (stddev: 0.00023294672175019573)	1061.1266799026703 iter/sec (stddev: 0.00018060320885434483)	1.43
benchmarks/test_guard_lifecycle.py::test_g6_build_child_container	616922.6605220381 iter/sec (stddev: 3.715201003928183e-7)	1126700.927576428 iter/sec (stddev: 4.1378652249439554e-8)	1.83
benchmarks/test_guard_lifecycle.py::test_g6b_build_child_container_auto_scope	734411.3929540286 iter/sec (stddev: 6.08801952519743e-8)	1036355.9899417069 iter/sec (stddev: 4.3060571714441033e-8)	1.41
benchmarks/test_guard_lifecycle.py::test_g7_request_lifecycle_batch	2361.049484052426 iter/sec (stddev: 0.00004888846724468825)	3271.4591629564534 iter/sec (stddev: 0.000009229771135313224)	1.39
benchmarks/test_guard_lifecycle.py::test_g7c_event_loop_floor_control	57596.78863108535 iter/sec (stddev: 0.0000023023190759449053)	73054.24112724997 iter/sec (stddev: 0.0000013365268275422494)	1.27
benchmarks/test_guard_lifecycle.py::test_g13_teardown_at_scale	44583.38100972135 iter/sec (stddev: 0.0000021225133640635065)	61634.29912022505 iter/sec (stddev: 0.0000021808385227602092)	1.38
benchmarks/test_guard_lifecycle.py::test_g13b_teardown_at_scale_async_no_finalizers	563.4749020753834 iter/sec (stddev: 0.00003065388973762685)	802.014201917468 iter/sec (stddev: 0.0001425391488858678)	1.42
benchmarks/test_guard_resolve.py::test_g1_transient_resolve	2052739.074909094 iter/sec (stddev: 3.523406532503467e-8)	3008481.284814736 iter/sec (stddev: 2.920374869649033e-8)	1.47
benchmarks/test_guard_resolve.py::test_g2_cached_resolve	3369558.5588639006 iter/sec (stddev: 1.4341991136431386e-8)	4786923.732775733 iter/sec (stddev: 1.5389072634008654e-8)	1.42
benchmarks/test_guard_resolve.py::test_g3_deep_chain	715514.4570475413 iter/sec (stddev: 4.65692090648329e-8)	1056026.9557198184 iter/sec (stddev: 3.7584032949087816e-8)	1.48
benchmarks/test_guard_resolve.py::test_g4_wide_resolve	472012.68071708444 iter/sec (stddev: 3.123355397671735e-7)	626482.7279945917 iter/sec (stddev: 3.1358607686122413e-7)	1.33
benchmarks/test_guard_resolve.py::test_g5_cross_scope	1858708.0752233567 iter/sec (stddev: 3.069798280846045e-8)	2759412.545912089 iter/sec (stddev: 3.202354432896829e-8)	1.48
benchmarks/test_guard_resolve.py::test_g9_context_resolve	1053365.6084545543 iter/sec (stddev: 1.4624605878624907e-7)	1489595.639952772 iter/sec (stddev: 1.6483873603149457e-7)	1.41
benchmarks/test_guard_resolve.py::test_g12_override_active_resolve	719698.0233514238 iter/sec (stddev: 6.074731741991982e-8)	1057983.080100332 iter/sec (stddev: 4.047735683117108e-8)	1.47
benchmarks/test_guard_resolve.py::test_g18_alias_hop	2117560.612541916 iter/sec (stddev: 1.320455111131115e-8)	3365320.7066519074 iter/sec (stddev: 1.6545091511550612e-8)	1.59
benchmarks/test_guard_validate.py::test_g10_validate_deep_chain	23025.9282617151 iter/sec (stddev: 0.00002276332321588352)	33726.56294767888 iter/sec (stddev: 0.000020314821428495196)	1.46
benchmarks/test_guard_validate.py::test_g11_validate_wide	14026.163732545592 iter/sec (stddev: 0.000027786077301247497)	20075.791466350274 iter/sec (stddev: 0.000020561362561656457)	1.43

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