test: the build family's probes run one check per obligation

[CaffeinatedCoder]

What

C11, C17, C18 and C19 ran as one check named claim. So no listed obligation of theirs earned credit, and PR #49 reported every per-obligation seed of C17 and C19 missed (C18's and C11's were the same). Each probe now runs one Row.Check per obligation its row's axes select, named by the claim's obligation id and with its baseline, following C22 (#47).

Probe	Checks per row	Baseline
C17	(FaultKind)9: a-cast-naming-no-member, reported-where-written, plus constructor-accessibility-too at site=attribute-argument. default: default-and-zero, reported-where-written. valid: a-member-is-silent (new, below)	Docs(src/CodoMetis.TypeKit.Analyzers/README.md#cmtk0014)
C18	names-each-base-member, a-warning-and-generated, generated-equality-never-calls-the-base, on every row	Docs(Generators README#equality), Docs(…#build-errors), Twin (the wrapped ints' own equality)
C19	the attribute's obligation (interface-and-unsealed, value-object-and-closed-set, flags-enum-and-read-only-struct, a-pair-once-at-the-later-argument), plus methods-the-compiler-calls where the site is not call	Docs(docs/type-constraints.md#the-kind-of-a-type-argument) or #two-parameters-apart; #constrain-a-type-parameter for the methods
C11	the parse shape's obligation (implicit-result, ref-text, abstract-string-constructor), or parsing-where-it-can-be-parsed (new, below) for ordinary	Docs(CHANGELOG.md#2.1.0-fixed); Twin (the catalogue's top-level value object of the same kind and wrapped type)

How each check is kept to its own obligation:

• C17. There is one compilation per row, and each check judges only its share of it:
  • the FaultKind places (a-cast…, default-and-zero);
  • RequireParameterlessConstructor's argument (constructor-accessibility-too; under default-and-zero, default there is Public and must stay silent).
  • A value counts as reported if a report sits at it or at the syntax around it (Rz.Reported). Where the report sits is reported-where-written's question alone (Rz.WhereWritten: at the value's own span, and nothing outside every place).
• C18. The two consumer checks mark the row's shape axes themselves. The equality check runs this project's own in-process twins, so it reads base-member alone and is credited with no shape pairs it never ran. The row is read from the rows file's copy, as Decls.Key does.
• C19. Every argument, from the catalogue and from the row's shapes, is now also used at a written call and at an inferred call.
  • The attribute's check judges the calls. It reads site only where the site is call, so in a foreach row it is not credited with site=foreach.
  • methods-the-compiler-calls judges each use at the row's site against the same argument's inferred call in the same compilation.
  • So a misjudged kind fails the attribute's check alone, and an unchecked site fails the methods' check alone (see the seeds).
  • BuildRoslyn.cs gains Agree over a subset of marks, At, Around, Outside, Reported and WhereWritten. The old Agree delegates to the new one.

Obligation decisions

• C17, valid values: the new obligation a-member-is-silent.
  • Sentence, quoted from the README's CMTK0014 example: public static readonly ImportFault Duplicate = new(FaultKind.Conflict);, the member written beside the reported new(default). Axes value=[valid].
  • Why not fold it into a-cast-naming-no-member: a valid value's silence is evidence for neither the cast obligation nor the default one. A rule that reports every value, members included, would pass both. No seed had ever shown the 86 valid rows could fail.
  • Its seed: CMTK0014 judges an explicit cast by its form, not by the value it names, so (FaultKind)2 is reported.
  • The first seed (members compared by name) failed the seeded build, because the shipping code's own constants were then reported.
• C11, ordinary: the new obligation parsing-where-it-can-be-parsed.
  • 169 of C11's 241 rows have parse-shape=ordinary, which none of the three obligations selects. Leaving them as claim would have kept half the probe unconverted. Making them inapplicable would have uncovered every ordinary pair.
  • Sentence, from the same file's 2.0.0 Generators entry (what the 2.1.0 fix must leave to a type that parses): `IParsable`/`ISpanParsable`/`IUtf8SpanParsable`, formatting in the invariant culture. Axes parse-shape=[ordinary].
  • Its seed: the generated [TypeConverter] is attached to a top-level value object only.
  • The first seed (an enum given no parsing) failed the seeded build, since code elsewhere parses an enum value object.
• Inapplicability declared in claims.json (build: declared applicability: an obligation says where it does not apply, read by the universe, the covering generator, the gate and Row.Check #50): one, for parsing-where-it-can-be-parsed: shape.wrapped ∈ {Uri, enum, flags-enum, NodaTime.Instant} and text.entry=ISpanParsable, cited not-promised:no-span-parsing-without-the-underlying-types (the Finding, below). Row.Check applies it, and it stays out of code because it depends on the axes alone. No other converted check decides inapplicability, so nothing else moved into the claims.

Outcome note keys and uncited excuses

• Note keys. Neither the four probes nor the build harness (BuildConsumer, BuildDecls, BuildSource, BuildRoslyn) called an inapplicable, expected-refusal or inconclusive key. None was converted, and no NoteKeys member was touched. Pass keys stay c.Note(NoteKeys.Compared, …).
• Not-promised entries added: one. C11's no-span-parsing-without-the-underlying-types, from src/CodoMetis.TypeKit.Generators/README.md#what-is-generated: "A Uri, an enum and a NodaTime type have neither." Its axes are the four wrapped types and text.entry=ISpanParsable. The sentence comes from fix: a value object over string implements ISpanParsable, as string does #52's narrowed README.
• Uncited excuses: one.
  • C11's ordinary check let a value object over a type that parses lack ISpanParsable (entry is not "ISpanParsable"), citing nothing.
  • The Generators README's "What is generated" table promises IParsable, ISpanParsable and IUtf8SpanParsable "when the underlying type can be parsed". No record or fact narrows that.
  • The exemption is gone, and the check runs. It failed 5 rows: see "Finding" below, where the fix (fix: a value object over string implements ISpanParsable, as string does #52) and the not-promised entry settle them.

Finding: value objects over a parseable type lack ISpanParsable (C11)

The rows. Probe C11ParseShapesBuildProbe, check parsing-where-it-can-be-parsed (Baseline.Twin), category violation. All five have parse-shape=ordinary and text.entry=ISpanParsable:

Row	shape.wrapped	shape.kind	shape.site	shape.inheritance	process.converter-lookup	text.culture
#174	Uri	validated-class	nested-in-class	base-depth-3	wrapped-type-converter-looked-up-first	invariant
#180	enum	record-struct	nested-two-levels	explicit-interface-member	wrapped-type-converter-looked-up-first	globalization-invariant
#183	flags-enum	validated-class	nested-in-struct	base-depth-3	none	invariant
#189	NodaTime.Instant	validated-class	nested-two-levels	explicit-interface-member	none	invariant
#205	string	sealed-record-class	nested-in-struct	base-depth-1	wrapped-type-converter-looked-up-first	tr-TR

The failure, for #205: violation:ISpanParsable: the value object over string, which parses, has no ISpanParsable. The other four read the same, with their wrapped type. Each row fails 3 of 3 times at HEAD (prober try --rows 174,180,183,189,205).

Why now. The ordinary check's presence test exempted ISpanParsable (entry is not "ISpanParsable") and cited nothing. Rule 2 makes that an uncited excuse: it is gone, and the check runs. The comparison with the catalogue still agrees, since both sides lack it; only the presence test fails.

Triage (.claude/skills/probe-triage):

1. A harness error? No. The check reads the value object's own interfaces by reflection (TextOps.Implements(ISpanParsable<>, Vo)). The catalogue's top-level value object over the same type lacks it too, so it is no artefact of the row's shape.
2. A fact? No. docs-query find ISpanParsable --kind fact finds nothing, and this is TypeKit's own generation, not a framework's.
3. Not promised? No. C11 has no not-promised entry. The only docs sentence is the Generators README's "What is generated" row, which promises it: "IParsable, ISpanParsable, IUtf8SpanParsable | when the underlying type can be parsed". design/aspects.md explains only the enum strategy, and no record narrows it.
4. A finding. ValueObjectParsableAspect.ResolveStrategy gives ISpanParsable only to its SpanParsable strategy.
   • string returns its own String strategy first, although string is ISpanParsable<string>. That breaks even the aspect's doc comment ("where the wrapped type has them").
   • Uri, the enums and NodaTime.Instant take the Uri, Enum and TypeConverter strategies, which generate IParsable alone.
   • No code-analyst agent was available on the host, so nothing was filed and no Skip was written.

The two sides, as the row's value object / the catalogue's / the wrapped type:

Wrapped type	Row's value object	Catalogue's value object	Wrapped type itself
string	no ISpanParsable	no ISpanParsable	is ISpanParsable<string>
Uri, enum, flags-enum, NodaTime.Instant	no ISpanParsable	no ISpanParsable	not ISpanParsable

Every one of them gets IParsable.

The fix landed as #52 (decision (a), a fix: with Found-By: PR), merged into release/2.2.0 at 0805ac5 and into this branch at f9ee5b2:

• a value object over string implements ISpanParsable;
• the Generators README's "What is generated" row is split: ISpanParsable and IUtf8SpanParsable are generated "where the underlying type has them", and the note says "A Uri, an enum and a NodaTime type have neither.";
• the Generators tests come with it (SpanParsingTests).

How C11 cites it (da2c291):

• C11 gains the not-promised entry no-span-parsing-without-the-underlying-types, quoting that sentence verbatim from src/CodoMetis.TypeKit.Generators/README.md#what-is-generated. Its axes are shape.wrapped ∈ {Uri, enum, flags-enum, NodaTime.Instant} and text.entry=ISpanParsable. universe (Axes.Build) accepts the quote.
• parsing-where-it-can-be-parsed declares itself inapplicable there (build: declared applicability: an obligation says where it does not apply, read by the universe, the covering generator, the gate and Row.Check #50's inapplicable, with the same axes and cite: not-promised:no-span-parsing-without-the-underlying-types), so Row.Check records those rows inapplicable before the body runs.
• The declaration includes text.entry, so the same types' Parse, TryParse, IParsable and type converter checks still run. string is not in it: its row now passes on the fix, not on an excuse.

Rows #174, #180, #183 and #189 are inapplicable, and #205 passes. The recorded reason, for #174: inapplicable:not-promised:no-span-parsing-without-the-underlying-types(declared shape.wrapped=Uri text.entry=ISpanParsable). C11 ended at inapplicable 4, pass 237, no violation, in prober try and in replay --family text at da2c291.

The exclusion and the rows (4d01780, 6e6beaa):

• With the entry alone, 11 of C11's 1048 pairs were uncovered (1037/1048).
  • 4 were the declared pairs, each of the four types × text.entry=ISpanParsable. They still counted as wanted, because C11's three parse-shape obligations want every pair they do not declare, though they never read shape.wrapped.
  • 7 were covered only by the 4 rows now inapplicable.
• The 4 pairs are excluded in claims.json (decision 102's form). The reason: "A Uri, an enum and a NodaTime type have neither (Generators README, What is generated: no ISpanParsable or IUtf8SpanParsable where the underlying type has none, not-promised no-span-parsing-without-the-underlying-types), so parsing-where-it-can-be-parsed is inapplicable there and no check that reads shape.wrapped can credit them." The universe goes from 1048 to 1044 pairs (242179 to 242175 in all).
• The rows: the 241 of da2c291 plus 6. I did not regenerate them whole:
  • prober rows --probe regenerated C11's rows whole as 157 rows (4d01780), which covered only 888 of 1044. The covering array puts a shape.wrapped value on a parse-shape row, which never reads it, and counts that row's wrapped pairs as covered. Where an axis has no unblocked value, it places an excluded pair anyway.
  • So C11 keeps its 241 rows, and gains the arrays prober rows --claim printed over the pairs those left: --seed 4, five ordinary rows over the 7, then --seed 1, one more. Row 243 of the first array was a record-struct with base-depth-3, which C11's declarations substitute, so it missed (base-depth-3, Uri).
  • The 247 rows were staged as one file and published.
  • C11Shapes.g.cs was generated again with decls.cs (GeneratedHarnessTests passes).
• C11's pairs: before 1048/1048. With the entry alone: 1037/1048. Now: 1044/1044, the 4 excluded pairs left out. C11 ends at inapplicable 4, pass 243, no violation.
• The prober side (an obligation's own axes narrowing the pairs it wants, and a covering array that never places an excluded pair) is for WS2's closing record, not this branch.

Before and after

• Before: replay --family build and --family text of release/2.2.0 at cfc9a2a, from an empty state, in an untouched worktree.
• After: the build family on this branch at 073f41e, which has build: declared applicability: an obligation says where it does not apply, read by the universe, the covering generator, the gate and Row.Check #50 merged in, and the text family at 6e6beaa, after fix: a value object over string implements ISpanParsable, as string does #52, the not-promised entry, the exclusion and C11's 247 rows. Formerly: the text family at da2c291, after fix: a value object over string implements ISpanParsable, as string does #52 and the not-promised entry.
• The other probes of the two families are listed as the control: unchanged, row for row.

Probe	before: rows	before: checks	after: rows	after: checks	after: checks by obligation
C11ParseShapesBuildProbe	pass 241	pass 241	inapplicable 4, pass 243	inapplicable 4, pass 243	abstract-string-constructor pass 22; implicit-result pass 28; parsing-where-it-can-be-parsed inapplicable 4, pass 171; ref-text pass 22
C17Cmtk0014Probe	pass 283	pass 283	pass 283	pass 512	a-cast-naming-no-member pass 75; a-member-is-silent pass 86; constructor-accessibility-too pass 32; default-and-zero pass 122; reported-where-written pass 197
C18Cmtk1023Probe	pass 280	pass 280	pass 280	pass 840	a-warning-and-generated pass 280; generated-equality-never-calls-the-base pass 280; names-each-base-member pass 280
C19Cmtk0010KindsProbe	pass 282	pass 282	pass 282	pass 508	a-pair-once-at-the-later-argument pass 80; flags-enum-and-read-only-struct pass 72; interface-and-unsealed pass 66; methods-the-compiler-calls pass 226; value-object-and-closed-set pass 64
C22GeneratorRefusalsProbe	pass 144	pass 432	pass 144	pass 432	abstract pass 20; closed-set-named-as-generated-member pass 4; equality-operator pass 20; every-error-links-the-table pass 144; generic-wrapped pass 24; named-as-generated-member pass 22; no-type-half-generated pass 144; over-object pass 6; over-system-enum pass 7; over-value-object pass 20; unsealed-record-class pass 21
C09BaseRecordEqualityProbe	pass 101	pass 101	pass 101	pass 101	claim pass 101
C20ShouldHoldProbe	pass 261	pass 261	pass 261	pass 261	claim pass 261
L05EqualityProbe	pass 309	pass 309	pass 309	pass 309	claim pass 309
C02OptionalTextProbe	pass 46	pass 46	pass 46	pass 46	claim pass 46
L03OneRuleProbe	expected-refusal 5, inapplicable 13, pass 568	expected-refusal 5, inapplicable 13, pass 568	expected-refusal 5, inapplicable 13, pass 568	expected-refusal 5, inapplicable 13, pass 568	claim expected-refusal 5, inapplicable 13, pass 568
L04RoundTripProbe	expected-refusal 78, inapplicable 127, pass 381	expected-refusal 78, inapplicable 127, pass 381	expected-refusal 78, inapplicable 127, pass 381	expected-refusal 78, inapplicable 127, pass 381	claim expected-refusal 78, inapplicable 127, pass 381

• Build family: 7 passed, 0 failed. Text family: 4 passed, 0 failed at 6e6beaa (and at da2c291). Before the fix it was 3 passed, 1 failed (C11, the 5 rows of the finding).
• No row that passed before fails, apart from those 5. No new violation elsewhere. C17, C18 and C19 end every row as before (pass 283, 280 and 282), now with 512, 840 and 508 passing checks.

prober status (after universe; 44 claims, 242179 pairs before, 242175 after C11's exclusion):

Claim	before	after
C11-parse-shapes-build	1048/1048, obligations 3, 3 uncredited	1044/1044 (4 pairs excluded, see the Finding), obligations 4, each seeded, 0 uncredited
C17-cmtk0014	1791/1832, obligations 4, 4 uncredited	1791/1832, obligations 5, each seeded, 0 uncredited
C18-cmtk1023	1067/1067, obligations 3, 3 uncredited	1067/1067, 0 uncredited
C19-cmtk0010-kinds	2242/2292, obligations 5, 5 uncredited	2242/2292, 0 uncredited
C22-generator-refusals	0/621, 0 uncredited	0/621, 0 uncredited

C19 keeps every pair, although its attribute checks no longer read a non-call site, because its methods' checks earn those pairs.

Seeds

prober seeds --probe P for the four probes and C22 as the control, one at a time, after the merge. 28 of 28 are caught, each by the check of its own obligation, named in the detail:

prober seeds: test/ws2-build-checks at 073f41e: 5 seeds of 1 probes
  C17Cmtk0014Probe/a-cast-naming-no-member (C17-cmtk0014): caught, 75 of 283 rows violate; e.g. #4 a-cast-naming-no-member: violation:p0 '(FaultKind)9': reported 0 times, expected once; p1 '(FaultKind)9': reported 0 times, expected once; p2 '(F…
  C17Cmtk0014Probe/default-and-zero (C17-cmtk0014): caught, 122 of 283 rows violate; e.g. #0 default-and-zero: violation:p0 '(FaultKind)0': reported 0 times, expected once; p1 '(FaultKind)0': reported 0 times, expected once; p2 '(F…
  C17Cmtk0014Probe/constructor-accessibility-too (C17-cmtk0014): caught, 32 of 283 rows violate; e.g. #4 constructor-accessibility-too: violation:ca3 '(ConstructorAccessibility)9': reported 0 times, expected once…
  C17Cmtk0014Probe/reported-where-written (C17-cmtk0014): caught, 139 of 283 rows violate; e.g. #0 reported-where-written: violation:p0 '(FaultKind)0': reported at Row0.cs:[625..640) '=> (FaultKind)0', not at the value; p1 '(FaultKind)0': repo…
  C17Cmtk0014Probe/a-member-is-silent (C17-cmtk0014): caught, 43 of 283 rows violate; e.g. #3 a-member-is-silent: violation:p0 '(FaultKind)2': reported, expected silent ((FaultKind)2 names no member of FaultKind: a host refuses a faul…
prober seeds: 5 caught, 374.9 s in all
prober seeds: test/ws2-build-checks at 073f41e: 3 seeds of 1 probes
  C18Cmtk1023Probe/names-each-base-member (C18-cmtk1023): caught, 130 of 280 rows violate; e.g. #0 names-each-base-member: violation:CMTK1023 reported 0 times in R0.cs, expected once…
  C18Cmtk1023Probe/a-warning-and-generated (C18-cmtk1023): caught, 280 of 280 rows violate; e.g. #0 a-warning-and-generated: violation:no CMTK1023 warning: the build reports none for the value object…
  C18Cmtk1023Probe/generated-equality-never-calls-the-base (C18-cmtk1023): caught, 83 of 280 rows violate; e.g. #4 generated-equality-never-calls-the-base: violation:over a base declaring GetHashCode by reference, two value objects holding 42 are unequal or hash apart: the ge…
prober seeds: 3 caught, 237.9 s in all
prober seeds: test/ws2-build-checks at 073f41e: 5 seeds of 1 probes
  C19Cmtk0010KindsProbe/methods-the-compiler-calls (C19-cmtk0010-kinds): caught, 59 of 282 rows violate; e.g. #6 methods-the-compiler-calls: violation:u7 '[x]': reported 0 times, expected once; u19 '[x]': reported 0 times, expected once; u35 '[x]': reported 0 t…
  C19Cmtk0010KindsProbe/interface-and-unsealed (C19-cmtk0010-kinds): caught, 32 of 282 rows violate; e.g. #1 interface-and-unsealed: violation:u0 'OrderHandler': reported 0 times, expected once; u1 'Runner.Run(a)': reported 0 times, expected once; u7 'D…
  C19Cmtk0010KindsProbe/value-object-and-closed-set (C19-cmtk0010-kinds): caught, 28 of 282 rows violate; e.g. #0 value-object-and-closed-set: violation:u0 'Guid': reported 0 times, expected once; u1 'Runner.Run(a)': reported 0 times, expected once; u19 'System.U…
  C19Cmtk0010KindsProbe/flags-enum-and-read-only-struct (C19-cmtk0010-kinds): caught, 33 of 282 rows violate; e.g. #2 flags-enum-and-read-only-struct: violation:u0 'Mutable': reported 0 times, expected once; u1 'Runner.Run(a)': reported 0 times, expected once…
  C19Cmtk0010KindsProbe/a-pair-once-at-the-later-argument (C19-cmtk0010-kinds): caught, 80 of 282 rows violate; e.g. #6 a-pair-once-at-the-later-argument: violation:u4 'Placed': reported 0 times, expected once; u13 'D6': reported 0 times, expected once; u29 'R6': reported 0 …
prober seeds: 5 caught, 487.0 s in all
prober seeds: test/ws2-build-checks at 4d01780: 4 seeds of 1 probes
  C11ParseShapesBuildProbe/implicit-result (C11-parse-shapes-build): caught, 28 of 247 rows violate; e.g. #2 implicit-result: violation:the value object over a implicit-result type has parsing: Parse, TryParse, IParsable, a type converter from st…
  C11ParseShapesBuildProbe/ref-text (C11-parse-shapes-build): caught, 22 of 247 rows violate; e.g. #4 ref-text: violation:the value object over a ref-text type has parsing: Parse, TryParse, IParsable, a type converter from string…
  C11ParseShapesBuildProbe/abstract-string-constructor (C11-parse-shapes-build): caught, 22 of 247 rows violate; e.g. #3 abstract-string-constructor: violation:the value object over a abstract-string-ctor type has parsing: Parse, TryParse, IParsable, a type converter fr…
  C11ParseShapesBuildProbe/parsing-where-it-can-be-parsed (C11-parse-shapes-build): caught, 16 of 247 rows violate; e.g. #37 parsing-where-it-can-be-parsed: violation:type-converter: the value object over char, which parses, has no type-converter…
prober seeds: 4 caught, 236.3 s in all
prober seeds: test/ws2-build-checks at 073f41e: 11 seeds of 1 probes
  unseeded build: 114.9 s, BUILD OK: 9 projects, 00:01:53.27
    C22GeneratorRefusalsProbe: 144 rows, 0 violations unseeded
  C22GeneratorRefusalsProbe/every-error-links-the-table (C22-generator-refusals): caught, 79 of 144 rows violate; e.g. #1 every-error-links-the-table: violation:these errors do not end with the link to the Build errors table: R1.cs CMTK1005 'O1.V1' cannot be generated as…
  C22GeneratorRefusalsProbe/unsealed-record-class (C22-generator-refusals): caught, 21 of 144 rows violate; e.g. #5 unsealed-record-class: violation:O5.V5 (unsealed-record-class (CMTK1006)) is not refused with CMTK1006 by name: its file reports no error…
  C22GeneratorRefusalsProbe/over-value-object (C22-generator-refusals): caught, 20 of 144 rows violate; e.g. #1 over-value-object: violation:O1.V1 (over-value-object (CMTK1005)) is not refused with CMTK1005 by name: its file reports no error…
  C22GeneratorRefusalsProbe/generic-wrapped (C22-generator-refusals): caught, 24 of 144 rows violate; e.g. #3 generic-wrapped: violation:O3.P3.V3 (generic-wrapped (CMTK1005)) is not refused with CMTK1005 by name: R3.cs LAMA0611 Error CS0305 in cod…
  C22GeneratorRefusalsProbe/abstract (C22-generator-refusals): caught, 20 of 144 rows violate; e.g. #4 abstract: violation:O4.P4.V4 (abstract (CMTK1005)) is not refused with CMTK1022 by name: its file reports no error…
  C22GeneratorRefusalsProbe/over-system-enum (C22-generator-refusals): caught, 7 of 144 rows violate; e.g. #8 over-system-enum: violation:V8 (over-System.Enum (CMTK1005)) is not refused with CMTK1005 by name: its file reports no error…
  C22GeneratorRefusalsProbe/named-as-generated-member (C22-generator-refusals): caught, 20 of 144 rows violate; e.g. #2 named-as-generated-member: violation:O2.TryFrom (named-as-generated-member (CMTK1005)) is not refused with CMTK1005 by name: R2.cs LAMA0611 Error C…
  C22GeneratorRefusalsProbe/closed-set-named-as-generated-member (C22-generator-refusals): caught, 4 of 144 rows violate; e.g. #11 closed-set-named-as-generated-member: violation:O11.P11.TryParse (closed-set-named-as-generated-member (CMTK1021)) is not refused with CMTK1021 by name: R11.c…
  C22GeneratorRefusalsProbe/equality-operator (C22-generator-refusals): caught, 20 of 144 rows violate; e.g. #0 equality-operator: violation:V0 (equality-operator (CMTK1011)) is not refused with CMTK1011 by name: R0.cs CS0111 Type 'V0' already defines…
  C22GeneratorRefusalsProbe/no-type-half-generated (C22-generator-refusals): caught, 79 of 144 rows violate; e.g. #1 no-type-half-generated: violation:O1.V1 (over-value-object (CMTK1005)) compiles: its file reports no error…
  C22GeneratorRefusalsProbe/over-object (C22-generator-refusals): caught, 6 of 144 rows violate; e.g. #15 over-object: violation:V15 (over-object (CMTK1005)) is not refused with CMTK1005 by name: its file reports no error…
prober seeds: 11 caught, 1355.0 s in all

• C11's run is at 4d01780 with the 247 rows of 6e6beaa in the checkout (the same run passes at da2c291 with 241 rows); the other four are at 073f41e.
• Before the merge (at 9ab1613), the same runs caught 26 of 28, each by its own check. The two first-draft seeds of the new obligations failed their seeded builds, and were replaced in 732baf2.
• In C19, the kind seeds fail the attribute's check, and the builder seed fails methods-the-compiler-calls.

Verdicts

Through build-verdict, at 6e6beaa:

• test --project test/CodoMetis.TypeKit.Prober.Tests: TESTS OK: 245 passed, 0 skipped.
• test --project test/CodoMetis.TypeKit.Conventions.Tests: TESTS OK: 1475 passed, 0 skipped, ProbeCiteTests and ProbeNoteKeyTests among them.
• prober register and publish ran SourceLint on each seed's registration.

Noticed on the way, not changed here

• C22 credits 0 of its 621 pairs. Its replay refuses an empty footprint ("the coverage report shows no executed file under src"), since the probe builds in a child process. C18 has watched paths for the same reason; C22 has none.
• C17's and C19's stored rows leave 41 and 50 pairs uncovered at replay (1791/1832, 2242/2292). test: C17 and C19 cover the five refused kinds C22 added #49 counted 0 from the stored rows: the replayed substitutions of excluded pairs credit other values.

Commits

• 9ab1613 test: the build family's probes run one check per obligation
• 732baf2 test: the two new obligations' seeds build
• 073f41e the merge of origin/release/2.2.0 (build: declared applicability: an obligation says where it does not apply, read by the universe, the covering generator, the gate and Row.Check #50)
• f9ee5b2 the merge of origin/release/2.2.0 (fix: a value object over string implements ISpanParsable, as string does #52)
• da2c291 test: C11 does not promise span parsing over a type that has none
• 4d01780 test: C11 excludes the span-parsing pairs no check can credit, and covers the rest again (its whole regeneration is replaced by the next)
• 6e6beaa test: C11 keeps its rows and adds the six that cover what the inapplicable four carried

🤖 Generated with Claude Code