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
Limited joints are charged the wrap-around distance · case 01
A branch that would require driving through a hard stop looks cheap and is selected.
Limited joints are charged the wrap-around distance · case 02
A branch that would require driving through a hard stop looks cheap and is selected.
Limited joints are charged the wrap-around distance · case 03
A branch that would require driving through a hard stop looks cheap and is selected.
Limited joints are charged the wrap-around distance · case 04
A branch that would require driving through a hard stop looks cheap and is selected.
Limited joints are charged the wrap-around distance · case 05
A branch that would require driving through a hard stop looks cheap and is selected.
Continuous joints are charged the long way around · case 01
A candidate 10 degrees away across the +/-180 seam costs as if it were 350 degrees away.
Continuous joints are charged the long way around · case 02
A candidate 10 degrees away across the +/-180 seam costs as if it were 350 degrees away.
Continuous joints are charged the long way around · case 03
A candidate 10 degrees away across the +/-180 seam costs as if it were 350 degrees away.
Continuous joints are charged the long way around · case 04
A candidate 10 degrees away across the +/-180 seam costs as if it were 350 degrees away.
Continuous joints are charged the long way around · case 05
A candidate 10 degrees away across the +/-180 seam costs as if it were 350 degrees away.
Branch cost ignores per-joint weights · case 01
Moving a heavy base joint is charged the same as moving a light wrist joint.
Branch cost ignores per-joint weights · case 02
Moving a heavy base joint is charged the same as moving a light wrist joint.
Branch cost ignores per-joint weights · case 03
Moving a heavy base joint is charged the same as moving a light wrist joint.
Branch cost ignores per-joint weights · case 04
Moving a heavy base joint is charged the same as moving a light wrist joint.
Branch cost ignores per-joint weights · case 05
Moving a heavy base joint is charged the same as moving a light wrist joint.
Equal-cost branches select the later candidate · case 01
Among symmetric solutions the last one listed is chosen, flipping the arm unexpectedly.
Equal-cost branches select the later candidate · case 02
Among symmetric solutions the last one listed is chosen, flipping the arm unexpectedly.
Equal-cost branches select the later candidate · case 03
Among symmetric solutions the last one listed is chosen, flipping the arm unexpectedly.
Equal-cost branches select the later candidate · case 04
Among symmetric solutions the last one listed is chosen, flipping the arm unexpectedly.
Equal-cost branches select the later candidate · case 05
Among symmetric solutions the last one listed is chosen, flipping the arm unexpectedly.
Selected branch index refers to the filtered list · case 01
When an infeasible candidate precedes the winner, the reported index points at the wrong solution.
Selected branch index refers to the filtered list · case 02
When an infeasible candidate precedes the winner, the reported index points at the wrong solution.
Selected branch index refers to the filtered list · case 03
When an infeasible candidate precedes the winner, the reported index points at the wrong solution.
Selected branch index refers to the filtered list · case 04
When an infeasible candidate precedes the winner, the reported index points at the wrong solution.
Selected branch index refers to the filtered list · case 05
When an infeasible candidate precedes the winner, the reported index points at the wrong solution.
SCARA quill extension is computed upward · case 01
Reachable heights are rejected and heights above the home plane are accepted.
SCARA quill extension is computed upward · case 02
Reachable heights are rejected and heights above the home plane are accepted.
SCARA quill extension is computed upward · case 03
Reachable heights are rejected and heights above the home plane are accepted.
SCARA quill extension is computed upward · case 04
Reachable heights are rejected and heights above the home plane are accepted.
SCARA quill extension is computed upward · case 05
Reachable heights are rejected and heights above the home plane are accepted.
SCARA rejects heights exactly at the quill end stops · case 01
Targets at full retraction or full extension report "out-of-stroke".
SCARA rejects heights exactly at the quill end stops · case 02
Targets at full retraction or full extension report "out-of-stroke".
SCARA rejects heights exactly at the quill end stops · case 03
Targets at full retraction or full extension report "out-of-stroke".
SCARA rejects heights exactly at the quill end stops · case 04
Targets at full retraction or full extension report "out-of-stroke".
SCARA rejects heights exactly at the quill end stops · case 05
Targets at full retraction or full extension report "out-of-stroke".
SCARA tool roll ignores the arm joints · case 01
The tool heading is off by the sum of the shoulder and elbow angles.
SCARA tool roll ignores the arm joints · case 02
The tool heading is off by the sum of the shoulder and elbow angles.
SCARA tool roll ignores the arm joints · case 03
The tool heading is off by the sum of the shoulder and elbow angles.
SCARA tool roll ignores the arm joints · case 04
The tool heading is off by the sum of the shoulder and elbow angles.
SCARA tool roll ignores the arm joints · case 05
The tool heading is off by the sum of the shoulder and elbow angles.
SCARA wrap subtracts one turn at most · case 01
Yaw requests of several turns leave the roll joint outside (-180,180].
SCARA wrap subtracts one turn at most · case 02
Yaw requests of several turns leave the roll joint outside (-180,180].
SCARA wrap subtracts one turn at most · case 03
Yaw requests of several turns leave the roll joint outside (-180,180].
SCARA wrap subtracts one turn at most · case 04
Yaw requests of several turns leave the roll joint outside (-180,180].
SCARA wrap subtracts one turn at most · case 05
Yaw requests of several turns leave the roll joint outside (-180,180].
SCARA law of cosines receives the radius instead of its square · case 01
Elbow angles are wrong everywhere except at unit radius.
SCARA law of cosines receives the radius instead of its square · case 02
Elbow angles are wrong everywhere except at unit radius.
SCARA law of cosines receives the radius instead of its square · case 03
Elbow angles are wrong everywhere except at unit radius.
SCARA law of cosines receives the radius instead of its square · case 04
Elbow angles are wrong everywhere except at unit radius.
SCARA law of cosines receives the radius instead of its square · case 05
Elbow angles are wrong everywhere except at unit radius.
SCARA reports reach failure before stroke failure · case 01
A target that is both too high and too far reports "unreachable" instead of "out-of-stroke".
SCARA reports reach failure before stroke failure · case 02
A target that is both too high and too far reports "unreachable" instead of "out-of-stroke".
SCARA reports reach failure before stroke failure · case 03
A target that is both too high and too far reports "unreachable" instead of "out-of-stroke".
SCARA reports reach failure before stroke failure · case 04
A target that is both too high and too far reports "unreachable" instead of "out-of-stroke".
SCARA reports reach failure before stroke failure · case 05
A target that is both too high and too far reports "unreachable" instead of "out-of-stroke".
Wrist singularity is detected only at an exact zero · case 01
Nearly aligned wrist axes produce huge, noise-driven a and g values instead of the documented convention.
Wrist singularity is detected only at an exact zero · case 02
Nearly aligned wrist axes produce huge, noise-driven a and g values instead of the documented convention.
Wrist singularity is detected only at an exact zero · case 03
Nearly aligned wrist axes produce huge, noise-driven a and g values instead of the documented convention.
Wrist singularity is detected only at an exact zero · case 04
Nearly aligned wrist axes produce huge, noise-driven a and g values instead of the documented convention.
Wrist singularity is detected only at an exact zero · case 05
Nearly aligned wrist axes produce huge, noise-driven a and g values instead of the documented convention.
Inverted wrist singularity reuses the upright roll formula · case 01
At b=180 the returned g has the wrong sign relation to the rotation.
Inverted wrist singularity reuses the upright roll formula · case 02
At b=180 the returned g has the wrong sign relation to the rotation.
Inverted wrist singularity reuses the upright roll formula · case 03
At b=180 the returned g has the wrong sign relation to the rotation.
Inverted wrist singularity reuses the upright roll formula · case 04
At b=180 the returned g has the wrong sign relation to the rotation.
Inverted wrist singularity reuses the upright roll formula · case 05
At b=180 the returned g has the wrong sign relation to the rotation.
Flipped wrist branch leaves the first angle unrotated · case 01
The flip solution reproduces a different orientation from the input matrix.
Flipped wrist branch leaves the first angle unrotated · case 02
The flip solution reproduces a different orientation from the input matrix.
Flipped wrist branch leaves the first angle unrotated · case 03
The flip solution reproduces a different orientation from the input matrix.
Flipped wrist branch leaves the first angle unrotated · case 04
The flip solution reproduces a different orientation from the input matrix.
Flipped wrist branch leaves the first angle unrotated · case 05
The flip solution reproduces a different orientation from the input matrix.
Wrist roll is taken from R20 without its minus sign · case 01
The final roll angle is mirrored to 180-g.
Wrist roll is taken from R20 without its minus sign · case 02
The final roll angle is mirrored to 180-g.
Wrist roll is taken from R20 without its minus sign · case 03
The final roll angle is mirrored to 180-g.
Wrist roll is taken from R20 without its minus sign · case 04
The final roll angle is mirrored to 180-g.
Wrist roll is taken from R20 without its minus sign · case 05
The final roll angle is mirrored to 180-g.
Wrist pitch is recovered with asin and cannot exceed 90 degrees · case 01
Tilts beyond 90 degrees fold back, giving the supplementary pitch.
Wrist pitch is recovered with asin and cannot exceed 90 degrees · case 02
Tilts beyond 90 degrees fold back, giving the supplementary pitch.
Wrist pitch is recovered with asin and cannot exceed 90 degrees · case 03
Tilts beyond 90 degrees fold back, giving the supplementary pitch.
Wrist pitch is recovered with asin and cannot exceed 90 degrees · case 04
Tilts beyond 90 degrees fold back, giving the supplementary pitch.
Wrist pitch is recovered with asin and cannot exceed 90 degrees · case 05
Tilts beyond 90 degrees fold back, giving the supplementary pitch.
First wrist angle is read from the bottom row instead of the last column · case 01
The first angle equals a function of the last angle, so a and g appear swapped.
First wrist angle is read from the bottom row instead of the last column · case 02
The first angle equals a function of the last angle, so a and g appear swapped.
First wrist angle is read from the bottom row instead of the last column · case 03
The first angle equals a function of the last angle, so a and g appear swapped.
First wrist angle is read from the bottom row instead of the last column · case 04
The first angle equals a function of the last angle, so a and g appear swapped.
First wrist angle is read from the bottom row instead of the last column · case 05
The first angle equals a function of the last angle, so a and g appear swapped.
Delta effector offset enlarges the tower radius · case 01
Every carriage height is computed for towers that are 2e too far out.
Delta effector offset enlarges the tower radius · case 02
Every carriage height is computed for towers that are 2e too far out.
Delta effector offset enlarges the tower radius · case 03
Every carriage height is computed for towers that are 2e too far out.
Delta effector offset enlarges the tower radius · case 04
Every carriage height is computed for towers that are 2e too far out.
Delta effector offset enlarges the tower radius · case 05
Every carriage height is computed for towers that are 2e too far out.
Delta towers are placed at 0, 120 and 240 degrees · case 01
Carriage heights belong to a rotated tower layout, so moves in y tilt the effector.
Delta towers are placed at 0, 120 and 240 degrees · case 02
Carriage heights belong to a rotated tower layout, so moves in y tilt the effector.
Delta towers are placed at 0, 120 and 240 degrees · case 03
Carriage heights belong to a rotated tower layout, so moves in y tilt the effector.
Delta towers are placed at 0, 120 and 240 degrees · case 04
Carriage heights belong to a rotated tower layout, so moves in y tilt the effector.
Delta towers are placed at 0, 120 and 240 degrees · case 05
Carriage heights belong to a rotated tower layout, so moves in y tilt the effector.
Delta rejects a point where one rod is exactly horizontal · case 01
A reachable point on the rod-length boundary returns "unreachable".
Delta rejects a point where one rod is exactly horizontal · case 02
A reachable point on the rod-length boundary returns "unreachable".
Delta rejects a point where one rod is exactly horizontal · case 03
A reachable point on the rod-length boundary returns "unreachable".
Delta rejects a point where one rod is exactly horizontal · case 04
A reachable point on the rod-length boundary returns "unreachable".
Delta rejects a point where one rod is exactly horizontal · case 05
A reachable point on the rod-length boundary returns "unreachable".
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 ↗