Unique-index vectors: pin holder identity by case (rows 23-24) #100

Merged
libby merged 2 commits from hands/variants-holder-case into main 2026-11-11 10:47:14 +00:00
4 changed files with 28 additions and 9 deletions

View file

@ -351,6 +351,7 @@ namespace GameCult.Caching.Tests
private static readonly Rec[] RepullBefore = Items_(("item:a1", "A"), ("item:b1", "C"));
private static readonly Rec[] RepullAfter = Items_(("item:a1", "A"), ("item:b1", "B"), ("item:b2", "B"), ("item:c1", "A"));
private static readonly Rec[] KeyCaseRecords = Items_(("a", "S"), ("z", "S"), ("B", "T"), ("C", "T"));
private static readonly Rec[] HolderCaseRecords = Items_(("a", "S"), ("A", "S"));
private static readonly Rec[] KeyPunctuationRecords = Items_(("_", "S"), ("a", "S"), ("Z", "T"), ("b", "T"));
private static readonly Rec[] SurrogateGroupRecords = Items_(("\uE000", "S"), ("\uE000b", "S"), ("\U0001F600", "T"), ("\U0001F600b", "T"));
@ -360,6 +361,7 @@ namespace GameCult.Caching.Tests
("repull-before.cc", RepullBefore), ("repull-after.cc", RepullAfter), ("alias-store.cc", AliasRecords),
("key-case-store.cc", KeyCaseRecords), ("key-punctuation-store.cc", KeyPunctuationRecords),
("surrogate-groups-store.cc", SurrogateGroupRecords),
("holder-case-store.cc", HolderCaseRecords),
};
private static readonly Refusal SurrogateRefusal = new("k\U0001F600", "k\uE000", Item, "Sku");
@ -368,6 +370,7 @@ namespace GameCult.Caching.Tests
private static readonly Refusal AliasRefusal = new("item:a", "item:b", Alias, "Sku");
private static readonly Refusal RepullRefusal = new("item:a1", "item:c1", Item, "Sku");
private static readonly Refusal KeyCaseRefusal = new("B", "C", Item, "Sku");
private static readonly Refusal HolderCaseRefusal = new("A", "a", Item, "Sku");
private static readonly Refusal KeyPunctuationRefusal = new("Z", "b", Item, "Sku");
private static readonly Refusal SurrogateGroupRefusal = new("\U0001F600", "\U0001F600b", Item, "Sku");
@ -402,6 +405,11 @@ namespace GameCult.Caching.Tests
new("load-key-case", new[] { Open("key-case-store.cc") }, KeyCaseRefusal),
new("load-key-punctuation", new[] { Open("key-punctuation-store.cc") }, KeyPunctuationRefusal),
new("load-surrogate-groups", new[] { Open("surrogate-groups-store.cc") }, SurrogateGroupRefusal),
// Rows 23-24: two holders of one value that differ only by case are two holders; a case-folding dedupe would see one.
new("batch-holders-by-case", new[] { Open(), Commit(HolderCaseRecords) }, HolderCaseRefusal),
new("load-holders-by-case", new[] { Open("holder-case-store.cc") }, HolderCaseRefusal),
// Row 25: the held `a` is replaced only by the exact key `a`, never by `A`.
new("batch-replaced-by-case", new[] { Open(), Commit(Items_(("a", "S"))), Commit(Items_(("A", "S"))) }, HolderCaseRefusal),
};
private static byte[] RefusalStoreBytes(Rec[] records)
@ -522,6 +530,7 @@ namespace GameCult.Caching.Tests
var count = reader.ReadArrayHeader();
Assert.That(count, Is.EqualTo(Cases.Length));
var walked = 0;
var failed = new List<string>();
for (var i = 0; i < count; i++)
{
Assert.That(reader.ReadArrayHeader(), Is.EqualTo(3));
@ -582,6 +591,10 @@ namespace GameCult.Caching.Tests
walked++;
}
catch (AssertionException failure)
{
failed.Add($"{label}: {failure.Message}");
}
finally
{
cache?.Dispose();
@ -589,7 +602,8 @@ namespace GameCult.Caching.Tests
}
}
Assert.That(walked, Is.EqualTo(22));
Assert.That(failed, Is.Empty, "rows that failed");
Assert.That(walked, Is.EqualTo(25));
}
}
}

View file

@ -19,7 +19,7 @@ reviewed commit. Other runtimes are held to the same canonical text and the same
`" "` and `item:d` holds Sku `U+0085`. Written through a role Sku.
- `blank-store.cc`: the same, only `item:c` and `item:d`.
- `pair-store.cc`, `surrogate-store.cc`, `types-store.cc`, `repull-before.cc`, `repull-after.cc`, `alias-store.cc`,
`key-case-store.cc`, `key-punctuation-store.cc`, `surrogate-groups-store.cc`: single-file stores written like
`key-case-store.cc`, `key-punctuation-store.cc`, `surrogate-groups-store.cc`, `holder-case-store.cc`: single-file stores written like
`duplicate-store.cc`, through roles, for the rows of `refusals.msgpack` that open or pull them. Schemas
`tests.unique_item`, `tests.uq_pair` (Sku and Alt both unique), `tests.uq_alias` (Sku and `alt` both unique),
`tests.uq_alpha` and `tests.uq_beta`, each slots `[0: Sku string, 1: second string]`, schema version the name plus
@ -51,19 +51,24 @@ last succeeds and the last refuses with exactly the RULE text built from those f
byte-identical. Schemas are those of the store files above; `second` is `Category` (plain) except in `tests.uq_pair`
and `tests.uq_alias`, where it is `Alt` (unique, alias `Alt` and `alt` respectively).
The 13 rows of the spec are a floor; rows 14-22 extend it with orderings that a code-point, case-insensitive or
The 13 rows of the spec are a floor; rows 14-25 extend it with orderings that a code-point, case-insensitive or
culture compare would flip. Rows, in file order: 1 batch-later-value-staged-first, 2 batch-surrogate-keys, 3
batch-interleaved-types, 4 batch-two-aliases, 5 held-then-landing, 6 held-in-the-least-group, 7 swap-off-a-held-value,
8 load-two-aliases, 9 load-surrogate-keys, 10 load-interleaved-types, 11 repull-held-pair, 12 load-repull-after, 13
commit-onto-a-loaded-holder, 14 batch-key-case (`B` before `a`), 15 batch-key-punctuation (`Z` before `_` before `a`),
16 batch-key-case-in-a-group, 17 batch-surrogate-groups (U+1F600 before U+E000, the reverse of code-point order), 18
batch-alias-case (alias `Sku` before `alt`), 19 load-alias-case, 20 load-key-case, 21 load-key-punctuation, 22
load-surrogate-groups. Rows 19-22 open a store holding the records of rows 18, 14, 15 and 17, so a runtime with no batch
load-surrogate-groups, 23 batch-holders-by-case (`a` and `A` holding one value: `A` before `a`), 24 load-holders-by-case, 25 batch-replaced-by-case (held `a`, then a commit of `A` on its value: `A` does not replace `a`).
Rows 19-22 and 24 open a store holding the records of rows 18, 14, 15, 17 and 23, so a runtime with no batch
commit still walks alias and cross-group key order.
A runtime performs every row its API can perform and asserts the count it walked:
A runtime performs every row its API can perform and asserts the count it walks. These are the counts a runtime must
assert, not a statement of which runtimes walk them today:
- C# and Rust: all 22.
- Python: 21, every row but 3 (`put_envelopes` takes one type).
- TypeScript: 11, rows 5, 8, 9, 10, 11, 12, 13, 19, 20, 21 and 22 (no batch commit, so no row 1-4, 6, 7 or 14-18;
`variants:idea:unique-index-runtimes.ts-batch-commit`).
- C# and Rust: all 25.
- Python: 24, every row but 3 (`put_envelopes` takes one type).
- TypeScript: 12, the rows that only open or pull a store: 5, 8, 9, 10, 11, 12, 13, 19, 20, 21, 22 and 24 (no batch commit, so no
row 1-4, 6, 7, 14-18, 23 or 25; `variants:idea:unique-index-runtimes.ts-batch-commit`).
Not pinned: the schema-name tiebreak. Record keys are unique across a whole cache, so two types never hold the same key and
two groups can never tie on alias and key pair; the final `schema` comparison cannot decide a refusal.

Binary file not shown.