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
Door placement between rooms: Even spans take the upper middle · case 01
Side doors are off-centre toward the bottom.
Door placement between rooms: Even spans take the upper middle · case 02
Side doors are off-centre toward the bottom.
Door placement between rooms: Even spans take the upper middle · case 03
Side doors are off-centre toward the bottom.
Door placement between rooms: Even spans take the upper middle · case 04
Side doors are off-centre toward the bottom.
Door placement between rooms: Even spans take the upper middle · case 05
Side doors are off-centre toward the bottom.
Door placement between rooms: Side door placed inside the neighbour room · case 01
Side doors are drawn one tile into the right room.
Door placement between rooms: Side door placed inside the neighbour room · case 02
Side doors are drawn one tile into the right room.
Door placement between rooms: Side door placed inside the neighbour room · case 03
Side doors are drawn one tile into the right room.
Door placement between rooms: Side door placed inside the neighbour room · case 04
Side doors are drawn one tile into the right room.
Door placement between rooms: Side door placed inside the neighbour room · case 05
Side doors are drawn one tile into the right room.
Door placement between rooms: Rooms listed right-to-left get no door · case 01
Connectivity depends on the order rooms were generated.
Door placement between rooms: Rooms listed right-to-left get no door · case 02
Connectivity depends on the order rooms were generated.
Door placement between rooms: Rooms listed right-to-left get no door · case 03
Connectivity depends on the order rooms were generated.
Door placement between rooms: Rooms listed right-to-left get no door · case 04
Connectivity depends on the order rooms were generated.
Door placement between rooms: Rooms listed right-to-left get no door · case 05
Connectivity depends on the order rooms were generated.
Door placement between rooms: Stacked rooms compared on the x axis · case 01
Vertically stacked rooms are only connected by coincidence.
Door placement between rooms: Stacked rooms compared on the x axis · case 02
Vertically stacked rooms are only connected by coincidence.
Door placement between rooms: Stacked rooms compared on the x axis · case 03
Vertically stacked rooms are only connected by coincidence.
Door placement between rooms: Stacked rooms compared on the x axis · case 04
Vertically stacked rooms are only connected by coincidence.
Door placement between rooms: Stacked rooms compared on the x axis · case 05
Vertically stacked rooms are only connected by coincidence.
Aligned stair placement: Stairs scanned column-major · case 01
Stair placement differs from the documented row-major rule.
Aligned stair placement: Stairs scanned column-major · case 02
Stair placement differs from the documented row-major rule.
Aligned stair placement: Stairs scanned column-major · case 03
Stair placement differs from the documented row-major rule.
Aligned stair placement: Stairs scanned column-major · case 04
Stair placement differs from the documented row-major rule.
Aligned stair placement: Stairs scanned column-major · case 05
Stair placement differs from the documented row-major rule.
Aligned stair placement: Stairs land in water below · case 01
The down stair leads onto a water tile.
Aligned stair placement: Stairs land in water below · case 02
The down stair leads onto a water tile.
Aligned stair placement: Stairs land in water below · case 03
The down stair leads onto a water tile.
Aligned stair placement: Stairs land in water below · case 04
The down stair leads onto a water tile.
Aligned stair placement: Stairs land in water below · case 05
The down stair leads onto a water tile.
Aligned stair placement: Entrance read as row, column · case 01
Stairs spawn next to the entrance on non-diagonal layouts.
Aligned stair placement: Entrance read as row, column · case 02
Stairs spawn next to the entrance on non-diagonal layouts.
Aligned stair placement: Entrance read as row, column · case 03
Stairs spawn next to the entrance on non-diagonal layouts.
Aligned stair placement: Entrance read as row, column · case 04
Stairs spawn next to the entrance on non-diagonal layouts.
Aligned stair placement: Entrance read as row, column · case 05
Stairs spawn next to the entrance on non-diagonal layouts.
Aligned stair placement: Exact minimum distance rejected · case 01
Small floors fail to place stairs.
Aligned stair placement: Exact minimum distance rejected · case 02
Small floors fail to place stairs.
Aligned stair placement: Exact minimum distance rejected · case 03
Small floors fail to place stairs.
Aligned stair placement: Exact minimum distance rejected · case 04
Small floors fail to place stairs.
Aligned stair placement: Exact minimum distance rejected · case 05
Small floors fail to place stairs.
Symplectic Euler substep drifts position with the pre-kick velocity · case 01
A stiff spring that should orbit with bounded energy gains amplitude every frame.
Symplectic Euler substep drifts position with the pre-kick velocity · case 02
A stiff spring that should orbit with bounded energy gains amplitude every frame.
Symplectic Euler substep drifts position with the pre-kick velocity · case 03
A stiff spring that should orbit with bounded energy gains amplitude every frame.
Symplectic Euler substep drifts position with the pre-kick velocity · case 04
A stiff spring that should orbit with bounded energy gains amplitude every frame.
Symplectic Euler substep drifts position with the pre-kick velocity · case 05
A stiff spring that should orbit with bounded energy gains amplitude every frame.
Spring restoring force pushes the mass away from rest · case 01
Displaced masses accelerate outward and the state grows exponentially.
Spring restoring force pushes the mass away from rest · case 02
Displaced masses accelerate outward and the state grows exponentially.
Spring restoring force pushes the mass away from rest · case 03
Displaced masses accelerate outward and the state grows exponentially.
Spring restoring force pushes the mass away from rest · case 04
Displaced masses accelerate outward and the state grows exponentially.
Spring restoring force pushes the mass away from rest · case 05
Displaced masses accelerate outward and the state grows exponentially.
Viscous damping force is applied as an acceleration without dividing by mass · case 01
Heavy bodies are over-damped and light ones under-damped relative to the contract.
Viscous damping force is applied as an acceleration without dividing by mass · case 02
Heavy bodies are over-damped and light ones under-damped relative to the contract.
Viscous damping force is applied as an acceleration without dividing by mass · case 03
Heavy bodies are over-damped and light ones under-damped relative to the contract.
Viscous damping force is applied as an acceleration without dividing by mass · case 04
Heavy bodies are over-damped and light ones under-damped relative to the contract.
Viscous damping force is applied as an acceleration without dividing by mass · case 05
Heavy bodies are over-damped and light ones under-damped relative to the contract.
Substep count truncates the frame to cell size ratio · case 01
Frames slightly longer than the 0.05 limit run one oversized substep and lose stability margin.
Substep count truncates the frame to cell size ratio · case 02
Frames slightly longer than the 0.05 limit run one oversized substep and lose stability margin.
Substep count truncates the frame to cell size ratio · case 03
Frames slightly longer than the 0.05 limit run one oversized substep and lose stability margin.
Substep count truncates the frame to cell size ratio · case 04
Frames slightly longer than the 0.05 limit run one oversized substep and lose stability margin.
Substep count truncates the frame to cell size ratio · case 05
Frames slightly longer than the 0.05 limit run one oversized substep and lose stability margin.
Every substep advances by the whole frame length · case 01
Frames split into several substeps advance simulated time by a multiple of dt.
Every substep advances by the whole frame length · case 02
Frames split into several substeps advance simulated time by a multiple of dt.
Every substep advances by the whole frame length · case 03
Frames split into several substeps advance simulated time by a multiple of dt.
Every substep advances by the whole frame length · case 04
Frames split into several substeps advance simulated time by a multiple of dt.
Every substep advances by the whole frame length · case 05
Frames split into several substeps advance simulated time by a multiple of dt.
Frame loop integrates one frame too many · case 01
Even zero requested frames move the body.
Frame loop integrates one frame too many · case 02
Even zero requested frames move the body.
Frame loop integrates one frame too many · case 03
Even zero requested frames move the body.
Frame loop integrates one frame too many · case 04
Even zero requested frames move the body.
Frame loop integrates one frame too many · case 05
Even zero requested frames move the body.
Verlet velocity kick uses only the start-of-step acceleration · case 01
Pendulum energy drifts steadily instead of oscillating around a constant.
Verlet velocity kick uses only the start-of-step acceleration · case 02
Pendulum energy drifts steadily instead of oscillating around a constant.
Verlet velocity kick uses only the start-of-step acceleration · case 03
Pendulum energy drifts steadily instead of oscillating around a constant.
Verlet velocity kick uses only the start-of-step acceleration · case 04
Pendulum energy drifts steadily instead of oscillating around a constant.
Verlet velocity kick uses only the start-of-step acceleration · case 05
Pendulum energy drifts steadily instead of oscillating around a constant.
Verlet drift applies the full acceleration dt squared term · case 01
Swing amplitude is systematically wrong at larger time steps.
Verlet drift applies the full acceleration dt squared term · case 02
Swing amplitude is systematically wrong at larger time steps.
Verlet drift applies the full acceleration dt squared term · case 03
Swing amplitude is systematically wrong at larger time steps.
Verlet drift applies the full acceleration dt squared term · case 04
Swing amplitude is systematically wrong at larger time steps.
Verlet drift applies the full acceleration dt squared term · case 05
Swing amplitude is systematically wrong at larger time steps.
Verlet never carries the new acceleration into the next step · case 01
After the first step the integrator keeps using the initial-angle acceleration.
Verlet never carries the new acceleration into the next step · case 02
After the first step the integrator keeps using the initial-angle acceleration.
Verlet never carries the new acceleration into the next step · case 03
After the first step the integrator keeps using the initial-angle acceleration.
Verlet never carries the new acceleration into the next step · case 04
After the first step the integrator keeps using the initial-angle acceleration.
Verlet never carries the new acceleration into the next step · case 05
After the first step the integrator keeps using the initial-angle acceleration.
Pendulum treats the degree input as radians · case 01
A 30 degree release swings as if released from 30 radians.
Pendulum treats the degree input as radians · case 02
A 30 degree release swings as if released from 30 radians.
Pendulum treats the degree input as radians · case 03
A 30 degree release swings as if released from 30 radians.
Pendulum treats the degree input as radians · case 04
A 30 degree release swings as if released from 30 radians.
Pendulum treats the degree input as radians · case 05
A 30 degree release swings as if released from 30 radians.
Pendulum force uses the small-angle linearization · case 01
Large releases oscillate with the harmonic period instead of the slower nonlinear period.
Pendulum force uses the small-angle linearization · case 02
Large releases oscillate with the harmonic period instead of the slower nonlinear period.
Pendulum force uses the small-angle linearization · case 03
Large releases oscillate with the harmonic period instead of the slower nonlinear period.
Pendulum force uses the small-angle linearization · case 04
Large releases oscillate with the harmonic period instead of the slower nonlinear period.
Pendulum force uses the small-angle linearization · case 05
Large releases oscillate with the harmonic period instead of the slower nonlinear period.
Pendulum potential energy uses cos theta instead of 1 - cos theta · case 01
The reported energy is offset and has the wrong sign convention at rest.
Pendulum potential energy uses cos theta instead of 1 - cos theta · case 02
The reported energy is offset and has the wrong sign convention at rest.
Pendulum potential energy uses cos theta instead of 1 - cos theta · case 03
The reported energy is offset and has the wrong sign convention at rest.
Pendulum potential energy uses cos theta instead of 1 - cos theta · case 04
The reported energy is offset and has the wrong sign convention at rest.
Pendulum potential energy uses cos theta instead of 1 - cos theta · case 05
The reported energy is offset and has the wrong sign convention at rest.
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 ↗