Understand the failure.
Verify the repair.
Small, reproducible software failures. The broken implementation, the fix that didn’t work, and the one that passed—preserved together.
Explore the cases ↓How results are verified ↗WHAT THE ARCHIVE CONTAINS
100840 executable cases. 20168 are open.
Every case records the implementation that fails, the fix that did not work, and the repair that passed its checks—with recorded outputs and source hashes. This release adds 100840 cases across 20168 failure mechanisms and 254 domains.
The open tier gives you the failure and the unsuccessful fix for one case in every mechanism. The remaining 80672 cases, 5 variants per mechanism, are member-only: the verified repair, its recorded checks, and the full fixture suite are held in the member archive. Read the methodology ↗
A RECORD OF WHAT WENT WRONG
Browse the archive / 100840
CTE lookup searches outer scopes before inner scopes · case 01
CTE lookup searches outer scopes before inner scopes.
CTE lookup searches outer scopes before inner scopes · case 02
CTE lookup searches outer scopes before inner scopes.
CTE lookup searches outer scopes before inner scopes · case 03
CTE lookup searches outer scopes before inner scopes.
CTE lookup searches outer scopes before inner scopes · case 04
CTE lookup searches outer scopes before inner scopes.
CTE lookup searches outer scopes before inner scopes · case 05
CTE lookup searches outer scopes before inner scopes.
Opening a query block overwrites its parent CTE dictionary · case 01
Opening a query block overwrites its parent CTE dictionary.
Opening a query block overwrites its parent CTE dictionary · case 02
Opening a query block overwrites its parent CTE dictionary.
Opening a query block overwrites its parent CTE dictionary · case 03
Opening a query block overwrites its parent CTE dictionary.
Opening a query block overwrites its parent CTE dictionary · case 04
Opening a query block overwrites its parent CTE dictionary.
Opening a query block overwrites its parent CTE dictionary · case 05
Opening a query block overwrites its parent CTE dictionary.
Leaving a query block retains its CTE definitions · case 01
Leaving a query block retains its CTE definitions.
Leaving a query block retains its CTE definitions · case 02
Leaving a query block retains its CTE definitions.
Leaving a query block retains its CTE definitions · case 03
Leaving a query block retains its CTE definitions.
Leaving a query block retains its CTE definitions · case 04
Leaving a query block retains its CTE definitions.
Leaving a query block retains its CTE definitions · case 05
Leaving a query block retains its CTE definitions.
An empty CTE relation is mistaken for an unresolved name · case 01
An empty CTE relation is mistaken for an unresolved name.
An empty CTE relation is mistaken for an unresolved name · case 02
An empty CTE relation is mistaken for an unresolved name.
An empty CTE relation is mistaken for an unresolved name · case 03
An empty CTE relation is mistaken for an unresolved name.
An empty CTE relation is mistaken for an unresolved name · case 04
An empty CTE relation is mistaken for an unresolved name.
An empty CTE relation is mistaken for an unresolved name · case 05
An empty CTE relation is mistaken for an unresolved name.
CTE lookup keeps scanning after finding an inner definition · case 01
CTE lookup keeps scanning after finding an inner definition.
CTE lookup keeps scanning after finding an inner definition · case 02
CTE lookup keeps scanning after finding an inner definition.
CTE lookup keeps scanning after finding an inner definition · case 03
CTE lookup keeps scanning after finding an inner definition.
CTE lookup keeps scanning after finding an inner definition · case 04
CTE lookup keeps scanning after finding an inner definition.
CTE lookup keeps scanning after finding an inner definition · case 05
CTE lookup keeps scanning after finding an inner definition.
An unresolved CTE name becomes an empty successful relation · case 01
An unresolved CTE name becomes an empty successful relation.
An unresolved CTE name becomes an empty successful relation · case 02
An unresolved CTE name becomes an empty successful relation.
An unresolved CTE name becomes an empty successful relation · case 03
An unresolved CTE name becomes an empty successful relation.
An unresolved CTE name becomes an empty successful relation · case 04
An unresolved CTE name becomes an empty successful relation.
An unresolved CTE name becomes an empty successful relation · case 05
An unresolved CTE name becomes an empty successful relation.
Session grouping compares every event with the first timestamp · case 01
Session grouping compares every event with the first timestamp.
Session grouping compares every event with the first timestamp · case 02
Session grouping compares every event with the first timestamp.
Session grouping compares every event with the first timestamp · case 03
Session grouping compares every event with the first timestamp.
Session grouping compares every event with the first timestamp · case 04
Session grouping compares every event with the first timestamp.
Session grouping compares every event with the first timestamp · case 05
Session grouping compares every event with the first timestamp.
Session grouping splits events separated by exactly the allowed gap · case 01
Session grouping splits events separated by exactly the allowed gap.
Session grouping splits events separated by exactly the allowed gap · case 02
Session grouping splits events separated by exactly the allowed gap.
Session grouping splits events separated by exactly the allowed gap · case 03
Session grouping splits events separated by exactly the allowed gap.
Session grouping splits events separated by exactly the allowed gap · case 04
Session grouping splits events separated by exactly the allowed gap.
Session grouping splits events separated by exactly the allowed gap · case 05
Session grouping splits events separated by exactly the allowed gap.
Session grouping includes events from other keys · case 01
Session grouping includes events from other keys.
Session grouping includes events from other keys · case 02
Session grouping includes events from other keys.
Session grouping includes events from other keys · case 03
Session grouping includes events from other keys.
Session grouping includes events from other keys · case 04
Session grouping includes events from other keys.
Session grouping includes events from other keys · case 05
Session grouping includes events from other keys.
Session grouping consumes source arrival order instead of event-time order · case 01
Session grouping consumes source arrival order instead of event-time order.
Session grouping consumes source arrival order instead of event-time order · case 02
Session grouping consumes source arrival order instead of event-time order.
Session grouping consumes source arrival order instead of event-time order · case 03
Session grouping consumes source arrival order instead of event-time order.
Session grouping consumes source arrival order instead of event-time order · case 04
Session grouping consumes source arrival order instead of event-time order.
Session grouping consumes source arrival order instead of event-time order · case 05
Session grouping consumes source arrival order instead of event-time order.
Session grouping overwrites the session start on every event · case 01
Session grouping overwrites the session start on every event.
Session grouping overwrites the session start on every event · case 02
Session grouping overwrites the session start on every event.
Session grouping overwrites the session start on every event · case 03
Session grouping overwrites the session start on every event.
Session grouping overwrites the session start on every event · case 04
Session grouping overwrites the session start on every event.
Session grouping overwrites the session start on every event · case 05
Session grouping overwrites the session start on every event.
Session grouping loses the final open session for each key · case 01
Session grouping loses the final open session for each key.
Session grouping loses the final open session for each key · case 02
Session grouping loses the final open session for each key.
Session grouping loses the final open session for each key · case 03
Session grouping loses the final open session for each key.
Session grouping loses the final open session for each key · case 04
Session grouping loses the final open session for each key.
Session grouping loses the final open session for each key · case 05
Session grouping loses the final open session for each key.
Watermark buffering rejects an event exactly at the watermark · case 01
Watermark buffering rejects an event exactly at the watermark.
Watermark buffering rejects an event exactly at the watermark · case 02
Watermark buffering rejects an event exactly at the watermark.
Watermark buffering rejects an event exactly at the watermark · case 03
Watermark buffering rejects an event exactly at the watermark.
Watermark buffering rejects an event exactly at the watermark · case 04
Watermark buffering rejects an event exactly at the watermark.
Watermark buffering rejects an event exactly at the watermark · case 05
Watermark buffering rejects an event exactly at the watermark.
A regressing watermark reopens already closed event time · case 01
A regressing watermark reopens already closed event time.
A regressing watermark reopens already closed event time · case 02
A regressing watermark reopens already closed event time.
A regressing watermark reopens already closed event time · case 03
A regressing watermark reopens already closed event time.
A regressing watermark reopens already closed event time · case 04
A regressing watermark reopens already closed event time.
A regressing watermark reopens already closed event time · case 05
A regressing watermark reopens already closed event time.
Watermark buffering releases events at the frontier prematurely · case 01
Watermark buffering releases events at the frontier prematurely.
Watermark buffering releases events at the frontier prematurely · case 02
Watermark buffering releases events at the frontier prematurely.
Watermark buffering releases events at the frontier prematurely · case 03
Watermark buffering releases events at the frontier prematurely.
Watermark buffering releases events at the frontier prematurely · case 04
Watermark buffering releases events at the frontier prematurely.
Watermark buffering releases events at the frontier prematurely · case 05
Watermark buffering releases events at the frontier prematurely.
Watermark release retains arrival order instead of event-time order · case 01
Watermark release retains arrival order instead of event-time order.
Watermark release retains arrival order instead of event-time order · case 02
Watermark release retains arrival order instead of event-time order.
Watermark release retains arrival order instead of event-time order · case 03
Watermark release retains arrival order instead of event-time order.
Watermark release retains arrival order instead of event-time order · case 04
Watermark release retains arrival order instead of event-time order.
Watermark release retains arrival order instead of event-time order · case 05
Watermark release retains arrival order instead of event-time order.
Watermark release leaves emitted events in the pending buffer · case 01
Watermark release leaves emitted events in the pending buffer.
Watermark release leaves emitted events in the pending buffer · case 02
Watermark release leaves emitted events in the pending buffer.
Watermark release leaves emitted events in the pending buffer · case 03
Watermark release leaves emitted events in the pending buffer.
Watermark release leaves emitted events in the pending buffer · case 04
Watermark release leaves emitted events in the pending buffer.
Watermark release leaves emitted events in the pending buffer · case 05
Watermark release leaves emitted events in the pending buffer.
Late events are reinserted into the pending buffer · case 01
Late events are reinserted into the pending buffer.
Late events are reinserted into the pending buffer · case 02
Late events are reinserted into the pending buffer.
Late events are reinserted into the pending buffer · case 03
Late events are reinserted into the pending buffer.
Late events are reinserted into the pending buffer · case 04
Late events are reinserted into the pending buffer.
Late events are reinserted into the pending buffer · case 05
Late events are reinserted into the pending buffer.
Top-k maintenance drops a deleted row without backfilling from the full relation · case 01
Top-k maintenance drops a deleted row without backfilling from the full relation.
Top-k maintenance drops a deleted row without backfilling from the full relation · case 02
Top-k maintenance drops a deleted row without backfilling from the full relation.
Top-k maintenance drops a deleted row without backfilling from the full relation · case 03
Top-k maintenance drops a deleted row without backfilling from the full relation.
Top-k maintenance drops a deleted row without backfilling from the full relation · case 04
Top-k maintenance drops a deleted row without backfilling from the full relation.
Top-k maintenance drops a deleted row without backfilling from the full relation · case 05
Top-k maintenance drops a deleted row without backfilling from the full relation.
Top-k maintenance treats score updates as insert-if-absent · case 01
Top-k maintenance treats score updates as insert-if-absent.
Top-k maintenance treats score updates as insert-if-absent · case 02
Top-k maintenance treats score updates as insert-if-absent.
Top-k maintenance treats score updates as insert-if-absent · case 03
Top-k maintenance treats score updates as insert-if-absent.
Top-k maintenance treats score updates as insert-if-absent · case 04
Top-k maintenance treats score updates as insert-if-absent.
Top-k maintenance treats score updates as insert-if-absent · case 05
Top-k maintenance treats score updates as insert-if-absent.
INSPECTABLE BY DESIGN
Every result has a runnable source.
Runnable implementations with recorded outputs, source hashes, and explicit contracts. Related variants share a failure mechanism and belong together in evaluation splits.
Read the methodology ↗