Runs
Every recorded run, newest first. A run groups every result that shares a date; targets not re-tested that day are carried forward. Open one for the full results.
-
Suite size954 tests
-
Suite on
Per-region scoring lands complete. 2.0.0-pre put the scoring logic in place, comparing each target against every region's recorded answer, but the evidence half was never wired: no test recorded what a target actually answered and the classifier never read one, so a fail could not be credited to a region the target matched and the score could only ever subtract. 2.0.0 closes that loop, and the seed split runs its whole lifecycle in the same release.
What changed:
- The split tests now record what the target actually answered
(
src/observation-sink.ts), in the same shape the registry stores each region's answer, and the classifier carries it onto the verdict. An engine that matches a rejecting region on a split is now scored as passing in that region. Evidence only ever redeems a committed fail: a pass keeps the committed assertion's deliberate wording tolerance and is never held to the byte-exact recorded string. Committed results predate the capture, so published scores move on each target's next run, not in this change. - Headline ties now prefer a region the registry characterises. A region absent from every split row ties the top score by having nothing recorded about it, and the Region column must not answer "conformant to what?" with a region the suite knows nothing about. Only the label is affected; a strictly higher score still wins whatever its source.
- The seeded
{ NULL: false }split closed its own loop. eu-west-2 and eu-central-1 were the last regions accepting it; by the 2026-07-17 sweep both reject it again, so every region agrees, the split is retired, and its test asserts the shared rejection. The detect, admit and reconcile path the pre-release introduced, exercised end to end within days. - A tooling test now asserts every tracked file is text, after a raw NUL byte used as a delimiter in one source file made grep classify the file as binary and silently skip it.
7 emulatorsSuite size954 tests - The split tests now record what the target actually answered
(
-
Suite size954 tests
-
Suite size954 tests
-
Suite size954 tests+81
-
Suite on
The scores barely move in this release, but what they mean has changed.
Until now the suite pinned one region, eu-west-2, as ground truth. That was quietly unfair: real DynamoDB disagrees with itself in a handful of places, and a one-region baseline takes a side without saying so. The clearest case is the
{ NULL: false }attribute value - accepted and normalised to{ NULL: true }in eu-west-2 and eu-central-1, rejected with a ValidationException in us-east-1 and ap-southeast-2. An engine matching us-east-1 on that behaviour was marked non-conformant for doing exactly what real DynamoDB does in Virginia. From 2.0.0, ground truth is per region.What changed:
- A weekly sweep runs the full suite against real DynamoDB in every
commercial region and publishes per-region ground truth
(
.github/workflows/sweep.yml,ground-truth/). It gates nothing: PR scoring stays offline and deterministic. - Confirmed regional splits live in a checked-in registry
(
registry/splits.json), each row carrying what every named region actually returned, when, and who admitted it. Detection is automatic; admission is not. The sweep files an issue with the evidence, and only a human commits a row. The registry ships seeded with the{ NULL: false }split. - Each target is scored against every observed region's expectations, and
its published number is its best-matching region - named in the results
table's new Region column, with the full per-region view in
results/summary.json(a versioned, additive artefact; the per-targetresults/<slug>.jsonfiles are unchanged in shape). - A third result state, indeterminate, for a failed observation: a timeout, an exhausted throttle, a transport fault. It is excluded from both sides of the score and cannot become a split, a registry row, or a fail. An absent answer is not a different answer.
- Region health is tracked in
registry/regions.json. A region that cannot complete a sweep is published as unresolved rather than silently omitted; two consecutive misses drop it from the scored set and page a maintainer in the same act.
One deliberate departure from the RFC that proposed this (#75): the RFC suggested a behaviour conforms if it matches any real region. 2.0.0 scores each target against one region at a time and headlines the best match, so a target only passes a behaviour when at least one real region does what it did, and its headline reflects one coherent region rather than a mix. Match-any scoring would have accepted an engine that combines eu-west-2's answer on one behaviour with us-east-1's on another - a deployment that exists nowhere. That is stricter than the RFC asked for, and it is deliberate.
No score moves at release: the one admitted split pins eu-west-2, which is the only region in the health record until the first sweep runs. Per-target deltas will be published once the sweep admits more regions; the expected movement is roughly a tenth of a percent for the six engines that match us-east-1 on the
{ NULL: false }split.The suite also grew to 954 tests, up 81, all characterised against real DynamoDB - the control-plane pins in eu-west-2, everything else across four regions (eu-west-2, eu-central-1, us-east-1, ap-southeast-2):
- UpdateTable AttributeDefinitions reconciliation (#77, #78, #79): delta-fed GSI adds merge into the stored union rather than replacing it, a conflicting redeclaration of an existing key keeps the stored type, and deleting a GSI prunes only its orphaned key attributes. An unused definition on an add is silently dropped where CreateTable rejects the same shape.
- ProjectionExpression validation (#81): duplicate paths, alias collisions and parent/child overlaps rejected identically on GetItem, Query, Scan and BatchGetItem, pinned with exact messages; legal shared-prefix projections guarded as accepted; GetItem's misfiled validation tests rehomed into Tier 3.
- Expression-size limit (#80): every expression parameter caps at 4096 bytes, measured on the raw string before ExpressionAttributeNames substitution - 4096 accepted, 4097 rejected, on all five expression surfaces. No tracked target enforces this limit today, so the Tier 3 movement it causes is new coverage, not a regression.
- Empty set members (#82): empty strings in an SS and zero-length members in a BS are accepted and round-trip intact through every write path, and contains() can find them; an empty NS member and duplicate empty members are rejected, with messages pinned. One existing assertion got stricter: the empty-binary round-trip now asserts byte length zero.
7 emulatorsSuite size873 tests - A weekly sweep runs the full suite against real DynamoDB in every
commercial region and publishes per-region ground truth
(
-
Suite size873 tests
-
Suite size873 tests
-
Suite on
Grew to 873 tests, up 49, all characterised against real DynamoDB in eu-west-2. New coverage in three areas:
- ConsumedCapacity: the transactional read/write split on a same-token replay and on ExecuteTransaction, and a correction that single-item operations report the aggregate CapacityUnits and omit the split, which is transactional-only.
- Empty-binary key values: rejected as a top-level ValidationException on every path, including secondary-index keys and inside transactions, mirroring the empty-string rejection.
- Expression, limit and response-shape parity: KeyConditionExpression operand and nested-path rules, ExpressionAttributeNames/Values hygiene, projection validation and list-index fidelity, reversed-bounds BETWEEN, read-path key-length and segment caps, whitespace numbers, batch unprocessed fields and cross-table projection mixing, the bare no-op upsert, multi-subpath UPDATED_NEW, filter operand ordering, hash-only-GSI pagination, and CreateTable spec validation.
7 emulatorsSuite size873 tests+49 -
Suite on
Grew to 824 tests, up 7, in two parts: two sibling-parity gaps where one half of a rule was pinned and the other was not, and the capacity accounting of conditional and idempotent transactional writes. All characterised against real DynamoDB in eu-west-2.
The first covers the LSI side of the INCLUDE-projection-without-NonKeyAttributes rejection; the GSI side already had it. An LSI declared with ProjectionType INCLUDE and no NonKeyAttributes is rejected as a ValidationException in tier1 and pinned to the exact message in tier3, the same wording the GSI case returns.
The second pins the Query message for Select SPECIFIC_ATTRIBUTES with no ProjectionExpression, which Scan already had. Query and Scan enforce the same rule but word it differently: Query wraps the phrase in the "1 validation error detected:" envelope, Scan returns it bare.
The transaction cases settle what a conditional TransactWriteItems actually costs. A passing condition adds no read capacity: a conditional write bills the same 2 WCU per sub-1KB item as an unconditional one, and a standalone ConditionCheck costs 2 WCU, billed as write not read. Idempotent replay splits the accounting - the first call reports 2 write capacity units, a same-token replay within the window reports 2 read capacity units for re-reading the stored result. A failing condition cancels the transaction, and the response carries no ConsumedCapacity at all. Answers #27.
7 emulatorsSuite size824 tests+7 -
Suite size817 tests
-
Suite size817 tests+55
-
Suite on
Grew to 762 tests, up 18, covering a malformed value in the lookup Key of a TransactWriteItems Update, Delete, or ConditionCheck - the path a Put item key does not take. Captured across four regions (eu-west-2, us-east-1, ap-southeast-2, eu-central-1), where every string was identical, so they pin exactly. An empty-string Key surfaces as a top-level ValidationException with the same message a Put item key gives; a wrong-typed or non-scalar Key cancels with a ValidationError reason carrying "The provided key element does not match the schema" - the key-only form, not the "Type mismatch for key" message the item-key path returns. The same run confirmed the BatchWriteItem table-key schema-mismatch message is region-invariant.
7 emulatorsSuite size762 tests+18 -
Suite on
Grew to 744 tests, up 8, pinning the Select / ProjectionExpression rules on Query and Scan. A ProjectionExpression is only valid with Select SPECIFIC_ATTRIBUTES, and ALL_PROJECTED_ATTRIBUTES is only valid with an IndexName; real DynamoDB rejects both with a ValidationException before reading anything. The cases span ALL_ATTRIBUTES, COUNT, and ALL_PROJECTED_ATTRIBUTES, including the request that breaks both rules at once, where AWS reports the ProjectionExpression one. They assert the contractual phrase, so they hold whether or not the engine carries the wrapper AWS adds on Query but not Scan.
7 emulatorsSuite size744 tests+8 -
Suite on
Grew to 736 tests, up 30, covering what TransactWriteItems and BatchWriteItem do with an item whose key value is malformed - the wrong type, non-scalar, or an empty string - across both table and index keys. PutItem already covered this; the transactional and batch paths covered none of it.
Characterising it against real AWS turned up a split worth pinning. An empty-string key value is rejected by up-front input validation, so even inside a transaction it surfaces as a top-level ValidationException. A wrong-typed or non-scalar key value is caught while the transaction runs, so it comes back as a TransactionCanceledException carrying a ValidationError reason rather than a top-level error. BatchWriteItem has no cancellation path, so every variant there is a plain ValidationException. The tests pin both halves, which catches two opposite mistakes: an engine that wraps the empty-string case as a cancellation, and one that surfaces the type-mismatch case as a top-level error.
7 emulatorsSuite size736 tests+30 -
Suite on
Grew to 706 tests, up 7, tightening secondary-index behaviour in Tier 1. Query and Scan on a GSI or LSI now assert sparse membership: an item that omits the index key stays off the index but remains on the base table. And PutItem now rejects an item whose GSI or LSI key value is the wrong type, non-scalar, or an empty string while the base-table keys are valid, holding an index key to its declared scalar type the same way a table key is held.
7 emulatorsSuite size706 tests+7 -
Suite size699 tests
-
Suite size699 tests
-
Suite size699 tests
-
Suite on
Real DynamoDB in eu-west-2 reworded a chunk of its validation errors, and the Tier 3 error-message tests moved with it. They now assert the contract the error carries - its type, the field it objects to, and the constraint - rather than the exact prose, because AWS varies the wrapper, the echoed input value and the field casing from one region to the next. A four-region capture in June found eu-west-2 and eu-central-1 on the new wording and us-east-1 and ap-southeast-2 still on the old, so the line between contract and cosmetic is drawn from what is invariant across all four.
This moves some Tier 3 numbers. Targets that were only ever marked down for wording DynamoDB itself renders inconsistently now pass those checks, so their Tier 3 scores rise: the suite has stopped counting a cosmetic difference as a behavioural one. Genuine behavioural divergences are still pinned exactly, for example PutItem with a { NULL: false } attribute, which DynamoDB now accepts in eu-west-2 and normalises to { NULL: true }.
7 emulatorsSuite size699 tests -
Suite size699 tests+15
-
Suite on
Grew to 684 tests with a control-plane and table-configuration sweep: the CreateTable/UpdateTable config parameters and the secondary control-plane operations (limits, backups and PITR, exports and imports, Kinesis, resource policies, contributor insights), each characterised against real AWS and probe-skipped where a target doesn't implement it.
The published percentage changed with it. It now measures correctness over implemented operations, Pass / (Pass + Fail), so skips no longer count against the score. A skip is honest scope; a fail is a bug.
7 emulatorsSuite size684 tests+29 -
Suite size655 tests+30
-
Suite on
Grew to 625 tests, up 24 on the previous run: eleven more in Tier 1 and thirteen more in Tier 3, tightening coverage of core operations and the strict edge cases.
7 emulatorsSuite size625 tests+24 -
7 emulatorsSuite size601 tests
-
Suite on
Grew to 601 tests, up 29, and every new test landed in Tier 3: a broader strict surface spanning validation ordering, exact error messages, service limits, and the legacy request shapes. Floci and Ministack were added to the run as targets the same day.
6 emulatorsSuite size601 tests+29 -
Suite size572 tests
-
Suite on
Grew to 572 tests, up 46 on the first run: thirty-six more in Tier 1 and ten more in Tier 2, deepening coverage of the core and complete-feature behaviour.
4 emulatorsSuite size572 tests+46 -
Suite size526 tests
-
Suite on
The suite was established with 526 tests across the three tiers (267 Tier 1, 93 Tier 2, 166 Tier 3), run against Dynalite, DynamoDB Local, Dynoxide, and LocalStack, with live AWS DynamoDB as the baseline.
4 emulatorsSuite size526 tests