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
Delta carriage is solved on the lower root · case 01
Carriages are placed below the effector plane.
Delta carriage is solved on the lower root · case 02
Carriages are placed below the effector plane.
Delta carriage is solved on the lower root · case 03
Carriages are placed below the effector plane.
Delta carriage is solved on the lower root · case 04
Carriages are placed below the effector plane.
Delta carriage is solved on the lower root · case 05
Carriages are placed below the effector plane.
Delta compares the effector height against carriage travel · case 01
Moves whose carriages exceed the top are accepted.
Delta compares the effector height against carriage travel · case 02
Moves whose carriages exceed the top are accepted.
Delta compares the effector height against carriage travel · case 03
Moves whose carriages exceed the top are accepted.
Delta compares the effector height against carriage travel · case 04
Moves whose carriages exceed the top are accepted.
Delta compares the effector height against carriage travel · case 05
Moves whose carriages exceed the top are accepted.
Delta rod horizontal distance uses the square of the summed offsets · case 01
Diagonal offsets produce wrong carriage heights while axis-aligned ones look correct.
Delta rod horizontal distance uses the square of the summed offsets · case 02
Diagonal offsets produce wrong carriage heights while axis-aligned ones look correct.
Delta rod horizontal distance uses the square of the summed offsets · case 03
Diagonal offsets produce wrong carriage heights while axis-aligned ones look correct.
Delta rod horizontal distance uses the square of the summed offsets · case 04
Diagonal offsets produce wrong carriage heights while axis-aligned ones look correct.
Delta rod horizontal distance uses the square of the summed offsets · case 05
Diagonal offsets produce wrong carriage heights while axis-aligned ones look correct.
Stewart platform composes roll, pitch and yaw in reverse order · case 01
Poses with more than one non-zero attitude angle produce wrong leg lengths.
Stewart platform composes roll, pitch and yaw in reverse order · case 02
Poses with more than one non-zero attitude angle produce wrong leg lengths.
Stewart platform composes roll, pitch and yaw in reverse order · case 03
Poses with more than one non-zero attitude angle produce wrong leg lengths.
Stewart platform composes roll, pitch and yaw in reverse order · case 04
Poses with more than one non-zero attitude angle produce wrong leg lengths.
Stewart platform composes roll, pitch and yaw in reverse order · case 05
Poses with more than one non-zero attitude angle produce wrong leg lengths.
Stewart pitch rotation matrix has transposed sine terms · case 01
Positive pitch tilts the platform the wrong way.
Stewart pitch rotation matrix has transposed sine terms · case 02
Positive pitch tilts the platform the wrong way.
Stewart pitch rotation matrix has transposed sine terms · case 03
Positive pitch tilts the platform the wrong way.
Stewart pitch rotation matrix has transposed sine terms · case 04
Positive pitch tilts the platform the wrong way.
Stewart pitch rotation matrix has transposed sine terms · case 05
Positive pitch tilts the platform the wrong way.
Stewart platform anchors are rotated by the inverse attitude · case 01
Legs are computed for the platform rotated the opposite way.
Stewart platform anchors are rotated by the inverse attitude · case 02
Legs are computed for the platform rotated the opposite way.
Stewart platform anchors are rotated by the inverse attitude · case 03
Legs are computed for the platform rotated the opposite way.
Stewart platform anchors are rotated by the inverse attitude · case 04
Legs are computed for the platform rotated the opposite way.
Stewart platform anchors are rotated by the inverse attitude · case 05
Legs are computed for the platform rotated the opposite way.
Stewart legs exactly at a stroke end are flagged · case 01
A pose whose legs sit exactly at lmin or lmax is reported as a limit violation.
Stewart legs exactly at a stroke end are flagged · case 02
A pose whose legs sit exactly at lmin or lmax is reported as a limit violation.
Stewart legs exactly at a stroke end are flagged · case 03
A pose whose legs sit exactly at lmin or lmax is reported as a limit violation.
Stewart legs exactly at a stroke end are flagged · case 04
A pose whose legs sit exactly at lmin or lmax is reported as a limit violation.
Stewart legs exactly at a stroke end are flagged · case 05
A pose whose legs sit exactly at lmin or lmax is reported as a limit violation.
Stewart limit report names the last violating leg · case 01
When several legs are out of range the report points at the wrong actuator.
Stewart limit report names the last violating leg · case 02
When several legs are out of range the report points at the wrong actuator.
Stewart limit report names the last violating leg · case 03
When several legs are out of range the report points at the wrong actuator.
Stewart limit report names the last violating leg · case 04
When several legs are out of range the report points at the wrong actuator.
Stewart limit report names the last violating leg · case 05
When several legs are out of range the report points at the wrong actuator.
Stewart pose vector is unpacked as yaw, pitch, roll · case 01
Roll and yaw commands are exchanged.
Stewart pose vector is unpacked as yaw, pitch, roll · case 02
Roll and yaw commands are exchanged.
Stewart pose vector is unpacked as yaw, pitch, roll · case 03
Roll and yaw commands are exchanged.
Stewart pose vector is unpacked as yaw, pitch, roll · case 04
Roll and yaw commands are exchanged.
Stewart pose vector is unpacked as yaw, pitch, roll · case 05
Roll and yaw commands are exchanged.
Hexapod plane distance ignores the coxa link · case 01
Feet land a coxa length short of the commanded position.
Hexapod plane distance ignores the coxa link · case 02
Feet land a coxa length short of the commanded position.
Hexapod plane distance ignores the coxa link · case 03
Feet land a coxa length short of the commanded position.
Hexapod plane distance ignores the coxa link · case 04
Feet land a coxa length short of the commanded position.
Hexapod plane distance ignores the coxa link · case 05
Feet land a coxa length short of the commanded position.
Hexapod femur pitch selects the knee-down triangle · case 01
The knee bends below the femur-foot line, dragging the tibia through the ground.
Hexapod femur pitch selects the knee-down triangle · case 02
The knee bends below the femur-foot line, dragging the tibia through the ground.
Hexapod femur pitch selects the knee-down triangle · case 03
The knee bends below the femur-foot line, dragging the tibia through the ground.
Hexapod femur pitch selects the knee-down triangle · case 04
The knee bends below the femur-foot line, dragging the tibia through the ground.
Hexapod femur pitch selects the knee-down triangle · case 05
The knee bends below the femur-foot line, dragging the tibia through the ground.
Hexapod tibia reports the interior knee angle · case 01
A straight leg reports 180 instead of 0 and bent legs are off by 180.
Hexapod tibia reports the interior knee angle · case 02
A straight leg reports 180 instead of 0 and bent legs are off by 180.
Hexapod tibia reports the interior knee angle · case 03
A straight leg reports 180 instead of 0 and bent legs are off by 180.
Hexapod tibia reports the interior knee angle · case 04
A straight leg reports 180 instead of 0 and bent legs are off by 180.
Hexapod tibia reports the interior knee angle · case 05
A straight leg reports 180 instead of 0 and bent legs are off by 180.
Hexapod femur triangle angle is taken at the foot · case 01
Femur pitch is wrong whenever femur and tibia differ in length.
Hexapod femur triangle angle is taken at the foot · case 02
Femur pitch is wrong whenever femur and tibia differ in length.
Hexapod femur triangle angle is taken at the foot · case 03
Femur pitch is wrong whenever femur and tibia differ in length.
Hexapod femur triangle angle is taken at the foot · case 04
Femur pitch is wrong whenever femur and tibia differ in length.
Hexapod femur triangle angle is taken at the foot · case 05
Femur pitch is wrong whenever femur and tibia differ in length.
Hexapod femur limits in degrees are compared with radians · case 01
Pitch limits almost never trigger, allowing poses beyond the servo range.
Hexapod femur limits in degrees are compared with radians · case 02
Pitch limits almost never trigger, allowing poses beyond the servo range.
Hexapod femur limits in degrees are compared with radians · case 03
Pitch limits almost never trigger, allowing poses beyond the servo range.
Hexapod femur limits in degrees are compared with radians · case 04
Pitch limits almost never trigger, allowing poses beyond the servo range.
Hexapod femur limits in degrees are compared with radians · case 05
Pitch limits almost never trigger, allowing poses beyond the servo range.
Hexapod reach test ignores the vertical component · case 01
Feet far below the body are accepted and then solved with a clamped cosine.
Hexapod reach test ignores the vertical component · case 02
Feet far below the body are accepted and then solved with a clamped cosine.
Hexapod reach test ignores the vertical component · case 03
Feet far below the body are accepted and then solved with a clamped cosine.
Hexapod reach test ignores the vertical component · case 04
Feet far below the body are accepted and then solved with a clamped cosine.
Hexapod reach test ignores the vertical component · case 05
Feet far below the body are accepted and then solved with a clamped cosine.
Redundant arm Jacobian sums links proximal to each joint · case 01
Joint velocities are computed for the wrong lever arms and the task error is not reduced.
Redundant arm Jacobian sums links proximal to each joint · case 02
Joint velocities are computed for the wrong lever arms and the task error is not reduced.
Redundant arm Jacobian sums links proximal to each joint · case 03
Joint velocities are computed for the wrong lever arms and the task error is not reduced.
Redundant arm Jacobian sums links proximal to each joint · case 04
Joint velocities are computed for the wrong lever arms and the task error is not reduced.
Redundant arm Jacobian sums links proximal to each joint · case 05
Joint velocities are computed for the wrong lever arms and the task error is not reduced.
Secondary motion is projected onto the task space · case 01
The joint-centering objective disturbs the end effector instead of moving in the self-motion manifold.
Secondary motion is projected onto the task space · case 02
The joint-centering objective disturbs the end effector instead of moving in the self-motion manifold.
Secondary motion is projected onto the task space · case 03
The joint-centering objective disturbs the end effector instead of moving in the self-motion manifold.
Secondary motion is projected onto the task space · case 04
The joint-centering objective disturbs the end effector instead of moving in the self-motion manifold.
Secondary motion is projected onto the task space · case 05
The joint-centering objective disturbs the end effector instead of moving in the self-motion manifold.
Joint-centering objective pushes joints away from centre · case 01
Self-motion increases the distance to the preferred configuration.
Joint-centering objective pushes joints away from centre · case 02
Self-motion increases the distance to the preferred configuration.
Joint-centering objective pushes joints away from centre · case 03
Self-motion increases the distance to the preferred configuration.
Joint-centering objective pushes joints away from centre · case 04
Self-motion increases the distance to the preferred configuration.
Joint-centering objective pushes joints away from centre · case 05
Self-motion increases the distance to the preferred configuration.
Pseudo-inverse uses a 2x2 inverse with the wrong off-diagonal sign · case 01
Task steps are skewed whenever the Jacobian rows are correlated.
Pseudo-inverse uses a 2x2 inverse with the wrong off-diagonal sign · case 02
Task steps are skewed whenever the Jacobian rows are correlated.
Pseudo-inverse uses a 2x2 inverse with the wrong off-diagonal sign · case 03
Task steps are skewed whenever the Jacobian rows are correlated.
Pseudo-inverse uses a 2x2 inverse with the wrong off-diagonal sign · case 04
Task steps are skewed whenever the Jacobian rows are correlated.
Pseudo-inverse uses a 2x2 inverse with the wrong off-diagonal sign · case 05
Task steps are skewed whenever the Jacobian rows are correlated.
Redundant arm forward kinematics uses relative angles as absolute · case 01
The task error is computed from the wrong tip position.
Redundant arm forward kinematics uses relative angles as absolute · case 02
The task error is computed from the wrong tip position.
Redundant arm forward kinematics uses relative angles as absolute · case 03
The task error is computed from the wrong tip position.
Redundant arm forward kinematics uses relative angles as absolute · case 04
The task error is computed from the wrong tip position.
Redundant arm forward kinematics uses relative angles as absolute · case 05
The task error is computed from the wrong tip position.
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 ↗