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
Clip mask cache: drop stale · case 01
Deleting a definition leaves cached samples visible under the deleted name.
Clip mask cache: drop stale · case 02
Deleting a definition leaves cached samples visible under the deleted name.
Clip mask cache: drop stale · case 03
Deleting a definition leaves cached samples visible under the deleted name.
Clip mask cache: drop stale · case 04
Deleting a definition leaves cached samples visible under the deleted name.
Clip mask cache: drop stale · case 05
Deleting a definition leaves cached samples visible under the deleted name.
Clip mask cache: clear stale · case 01
Clearing all masks retains valid-looking cache entries.
Clip mask cache: clear stale · case 02
Clearing all masks retains valid-looking cache entries.
Clip mask cache: clear stale · case 03
Clearing all masks retains valid-looking cache entries.
Clip mask cache: clear stale · case 04
Clearing all masks retains valid-looking cache entries.
Clip mask cache: clear stale · case 05
Clearing all masks retains valid-looking cache entries.
Clip mask cache: disabled cache result · case 01
Drawing with clipping disabled still uses cached coverage.
Clip mask cache: disabled cache result · case 02
Drawing with clipping disabled still uses cached coverage.
Clip mask cache: disabled cache result · case 03
Drawing with clipping disabled still uses cached coverage.
Clip mask cache: disabled cache result · case 04
Drawing with clipping disabled still uses cached coverage.
Clip mask cache: disabled cache result · case 05
Drawing with clipping disabled still uses cached coverage.
Clip mask cache: enable forgotten · case 01
Re-enabling leaves the cache bypass active.
Clip mask cache: enable forgotten · case 02
Re-enabling leaves the cache bypass active.
Clip mask cache: enable forgotten · case 03
Re-enabling leaves the cache bypass active.
Clip mask cache: enable forgotten · case 04
Re-enabling leaves the cache bypass active.
Clip mask cache: enable forgotten · case 05
Re-enabling leaves the cache bypass active.
Clip mask cache: missing definition full · case 01
A cache miss for an undefined mask populates full coverage.
Clip mask cache: missing definition full · case 02
A cache miss for an undefined mask populates full coverage.
Clip mask cache: missing definition full · case 03
A cache miss for an undefined mask populates full coverage.
Clip mask cache: missing definition full · case 04
A cache miss for an undefined mask populates full coverage.
Clip mask cache: missing definition full · case 05
A cache miss for an undefined mask populates full coverage.
Clip mask cache: revision order test · case 01
Cache invalidation compares revisions in the wrong direction.
Clip mask cache: revision order test · case 02
Cache invalidation compares revisions in the wrong direction.
Clip mask cache: revision order test · case 03
Cache invalidation compares revisions in the wrong direction.
Clip mask cache: revision order test · case 04
Cache invalidation compares revisions in the wrong direction.
Clip mask cache: revision order test · case 05
Cache invalidation compares revisions in the wrong direction.
Clip mask cache: definition accumulates · case 01
Replacing a mask definition merges coverage into the previous definition.
Clip mask cache: definition accumulates · case 02
Replacing a mask definition merges coverage into the previous definition.
Clip mask cache: definition accumulates · case 03
Replacing a mask definition merges coverage into the previous definition.
Clip mask cache: definition accumulates · case 04
Replacing a mask definition merges coverage into the previous definition.
Clip mask cache: definition accumulates · case 05
Replacing a mask definition merges coverage into the previous definition.
Clip mask cache: finite domain · case 01
A cached mask contains samples beyond the target extent.
Clip mask cache: finite domain · case 02
A cached mask contains samples beyond the target extent.
Clip mask cache: finite domain · case 03
A cached mask contains samples beyond the target extent.
Clip mask cache: finite domain · case 04
A cached mask contains samples beyond the target extent.
Clip mask cache: finite domain · case 05
A cached mask contains samples beyond the target extent.
Multisample clip resolve: rotation direction · case 01
Sample-mask rotation uses the opposite lane direction before lane enable filtering.
Multisample clip resolve: rotation direction · case 02
Sample-mask rotation uses the opposite lane direction before lane enable filtering.
Multisample clip resolve: rotation direction · case 03
Sample-mask rotation uses the opposite lane direction before lane enable filtering.
Multisample clip resolve: rotation direction · case 04
Sample-mask rotation uses the opposite lane direction before lane enable filtering.
Multisample clip resolve: rotation direction · case 05
Sample-mask rotation uses the opposite lane direction before lane enable filtering.
Multisample clip resolve: invert weight · case 01
Inversion clears all covered samples instead of complementing each sample bit.
Multisample clip resolve: invert weight · case 02
Inversion clears all covered samples instead of complementing each sample bit.
Multisample clip resolve: invert weight · case 03
Inversion clears all covered samples instead of complementing each sample bit.
Multisample clip resolve: invert weight · case 04
Inversion clears all covered samples instead of complementing each sample bit.
Multisample clip resolve: invert weight · case 05
Inversion clears all covered samples instead of complementing each sample bit.
Multisample clip resolve: enable domain · case 01
Lane enable indices are confused with Boolean coverage values.
Multisample clip resolve: enable domain · case 02
Lane enable indices are confused with Boolean coverage values.
Multisample clip resolve: enable domain · case 03
Lane enable indices are confused with Boolean coverage values.
Multisample clip resolve: enable domain · case 04
Lane enable indices are confused with Boolean coverage values.
Multisample clip resolve: enable domain · case 05
Lane enable indices are confused with Boolean coverage values.
Multisample clip resolve: empty vacuous · case 01
A clip with no enabled samples passes by vacuous all-sample logic.
Multisample clip resolve: empty vacuous · case 02
A clip with no enabled samples passes by vacuous all-sample logic.
Multisample clip resolve: empty vacuous · case 03
A clip with no enabled samples passes by vacuous all-sample logic.
Multisample clip resolve: empty vacuous · case 04
A clip with no enabled samples passes by vacuous all-sample logic.
Multisample clip resolve: empty vacuous · case 05
A clip with no enabled samples passes by vacuous all-sample logic.
Multisample clip resolve: weight denominator · case 01
Disabled sample weights still influence majority threshold.
Multisample clip resolve: weight denominator · case 02
Disabled sample weights still influence majority threshold.
Multisample clip resolve: weight denominator · case 03
Disabled sample weights still influence majority threshold.
Multisample clip resolve: weight denominator · case 04
Disabled sample weights still influence majority threshold.
Multisample clip resolve: weight denominator · case 05
Disabled sample weights still influence majority threshold.
Multisample clip resolve: weighted hit count · case 01
Weighted coverage uses the count of covered lanes rather than their weights.
Multisample clip resolve: weighted hit count · case 02
Weighted coverage uses the count of covered lanes rather than their weights.
Multisample clip resolve: weighted hit count · case 03
Weighted coverage uses the count of covered lanes rather than their weights.
Multisample clip resolve: weighted hit count · case 04
Weighted coverage uses the count of covered lanes rather than their weights.
Multisample clip resolve: weighted hit count · case 05
Weighted coverage uses the count of covered lanes rather than their weights.
Multisample clip resolve: all enabled count · case 01
All-sample resolve requires disabled lanes to be covered.
Multisample clip resolve: all enabled count · case 02
All-sample resolve requires disabled lanes to be covered.
Multisample clip resolve: all enabled count · case 03
All-sample resolve requires disabled lanes to be covered.
Multisample clip resolve: all enabled count · case 04
All-sample resolve requires disabled lanes to be covered.
Multisample clip resolve: all enabled count · case 05
All-sample resolve requires disabled lanes to be covered.
Multisample clip resolve: parity any · case 01
Parity resolve behaves as any-sample coverage on even overlaps.
Multisample clip resolve: parity any · case 02
Parity resolve behaves as any-sample coverage on even overlaps.
Multisample clip resolve: parity any · case 03
Parity resolve behaves as any-sample coverage on even overlaps.
Multisample clip resolve: parity any · case 04
Parity resolve behaves as any-sample coverage on even overlaps.
Multisample clip resolve: parity any · case 05
Parity resolve behaves as any-sample coverage on even overlaps.
Multisample clip resolve: strict tie · case 01
Strict weighted majority includes exact ties.
Multisample clip resolve: strict tie · case 02
Strict weighted majority includes exact ties.
Multisample clip resolve: strict tie · case 03
Strict weighted majority includes exact ties.
Multisample clip resolve: strict tie · case 04
Strict weighted majority includes exact ties.
Multisample clip resolve: strict tie · case 05
Strict weighted majority includes exact ties.
Multisample clip resolve: first order · case 01
First-sample resolve reads a disabled physical lane instead of the first enabled lane.
Multisample clip resolve: first order · case 02
First-sample resolve reads a disabled physical lane instead of the first enabled lane.
Multisample clip resolve: first order · case 03
First-sample resolve reads a disabled physical lane instead of the first enabled lane.
Multisample clip resolve: first order · case 04
First-sample resolve reads a disabled physical lane instead of the first enabled lane.
Multisample clip resolve: first order · case 05
First-sample resolve reads a disabled physical lane instead of the first enabled lane.
Stencil clip pass: coverage gate · case 01
Stencil operations run on every sample whenever incoming coverage is nonempty.
Stencil clip pass: coverage gate · case 02
Stencil operations run on every sample whenever incoming coverage is nonempty.
Stencil clip pass: coverage gate · case 03
Stencil operations run on every sample whenever incoming coverage is nonempty.
Stencil clip pass: coverage gate · case 04
Stencil operations run on every sample whenever incoming coverage is nonempty.
Stencil clip pass: coverage gate · case 05
Stencil operations run on every sample whenever incoming coverage is nonempty.
Stencil clip pass: reference mask · case 01
Stencil comparison masks stored bits but leaves reference metadata bits unmasked.
Stencil clip pass: reference mask · case 02
Stencil comparison masks stored bits but leaves reference metadata bits unmasked.
Stencil clip pass: reference mask · case 03
Stencil comparison masks stored bits but leaves reference metadata bits unmasked.
Stencil clip pass: reference mask · case 04
Stencil comparison masks stored bits but leaves reference metadata bits unmasked.
Stencil clip pass: reference mask · case 05
Stencil comparison masks stored bits but leaves reference metadata bits unmasked.
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 ↗