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
Inverse linear interpolation · case 01
The interval origin is not removed.
Inverse linear interpolation · case 02
The interval origin is not removed.
Inverse linear interpolation · case 03
The interval origin is not removed.
Inverse linear interpolation · case 04
The interval origin is not removed.
Inverse linear interpolation · case 05
The interval origin is not removed.
Linear extrapolation last slope · case 01
Last-value persistence discards the final slope.
Linear extrapolation last slope · case 02
Last-value persistence discards the final slope.
Linear extrapolation last slope · case 03
Last-value persistence discards the final slope.
Linear extrapolation last slope · case 04
Last-value persistence discards the final slope.
Linear extrapolation last slope · case 05
Last-value persistence discards the final slope.
Quadratic sequence extrapolation · case 01
Linear extrapolation omits second differences.
Quadratic sequence extrapolation · case 02
Linear extrapolation omits second differences.
Quadratic sequence extrapolation · case 03
Linear extrapolation omits second differences.
Quadratic sequence extrapolation · case 04
Linear extrapolation omits second differences.
Quadratic sequence extrapolation · case 05
Linear extrapolation omits second differences.
Finite impulse valid convolution · case 01
Unreversed kernel computes correlation instead of convolution.
Finite impulse valid convolution · case 02
Unreversed kernel computes correlation instead of convolution.
Finite impulse valid convolution · case 03
Unreversed kernel computes correlation instead of convolution.
Finite impulse valid convolution · case 04
Unreversed kernel computes correlation instead of convolution.
Finite impulse valid convolution · case 05
Unreversed kernel computes correlation instead of convolution.
Valid cross correlation · case 01
Kernel reversal changes correlation into convolution.
Valid cross correlation · case 02
Kernel reversal changes correlation into convolution.
Valid cross correlation · case 03
Kernel reversal changes correlation into convolution.
Valid cross correlation · case 04
Kernel reversal changes correlation into convolution.
Valid cross correlation · case 05
Kernel reversal changes correlation into convolution.
Discrete total variation · case 01
Net displacement cancels intermediate variation.
Discrete total variation · case 02
Net displacement cancels intermediate variation.
Discrete total variation · case 03
Net displacement cancels intermediate variation.
Discrete total variation · case 04
Net displacement cancels intermediate variation.
Discrete total variation · case 05
Net displacement cancels intermediate variation.
Piecewise linear slope sequence · case 01
The sample spacing is ignored.
Piecewise linear slope sequence · case 02
The sample spacing is ignored.
Piecewise linear slope sequence · case 03
The sample spacing is ignored.
Piecewise linear slope sequence · case 04
The sample spacing is ignored.
Piecewise linear slope sequence · case 05
The sample spacing is ignored.
Discrete laplacian periodic · case 01
The center stencil coefficient lacks its factor of two.
Discrete laplacian periodic · case 02
The center stencil coefficient lacks its factor of two.
Discrete laplacian periodic · case 03
The center stencil coefficient lacks its factor of two.
Discrete laplacian periodic · case 04
The center stencil coefficient lacks its factor of two.
Discrete laplacian periodic · case 05
The center stencil coefficient lacks its factor of two.
Binomial forward difference order three · case 01
The binomial stencil coefficients are replaced with unit alternation.
Binomial forward difference order three · case 02
The binomial stencil coefficients are replaced with unit alternation.
Binomial forward difference order three · case 03
The binomial stencil coefficients are replaced with unit alternation.
Binomial forward difference order three · case 04
The binomial stencil coefficients are replaced with unit alternation.
Binomial forward difference order three · case 05
The binomial stencil coefficients are replaced with unit alternation.
Newton step quadratic root · case 01
The reciprocal correction is returned without averaging the current iterate.
Newton step quadratic root · case 02
The reciprocal correction is returned without averaging the current iterate.
Newton step quadratic root · case 03
The reciprocal correction is returned without averaging the current iterate.
Newton step quadratic root · case 04
The reciprocal correction is returned without averaging the current iterate.
Newton step quadratic root · case 05
The reciprocal correction is returned without averaging the current iterate.
Secant root step rational · case 01
The interpolation correction omits the x span.
Secant root step rational · case 02
The interpolation correction omits the x span.
Secant root step rational · case 03
The interpolation correction omits the x span.
Secant root step rational · case 04
The interpolation correction omits the x span.
Secant root step rational · case 05
The interpolation correction omits the x span.
Newton divided difference two · case 01
Uniform-grid second differences are applied to nonuniform nodes.
Newton divided difference two · case 02
Uniform-grid second differences are applied to nonuniform nodes.
Newton divided difference two · case 03
Uniform-grid second differences are applied to nonuniform nodes.
Newton divided difference two · case 04
Uniform-grid second differences are applied to nonuniform nodes.
Newton divided difference two · case 05
Uniform-grid second differences are applied to nonuniform nodes.
Composite midpoint cell integral · case 01
The cell width is omitted.
Composite midpoint cell integral · case 02
The cell width is omitted.
Composite midpoint cell integral · case 03
The cell width is omitted.
Composite midpoint cell integral · case 04
The cell width is omitted.
Composite midpoint cell integral · case 05
The cell width is omitted.
Discrete energy squared norm · case 01
Squaring the total allows cancellation between samples.
Discrete energy squared norm · case 02
Squaring the total allows cancellation between samples.
Discrete energy squared norm · case 03
Squaring the total allows cancellation between samples.
Discrete energy squared norm · case 04
Squaring the total allows cancellation between samples.
Discrete energy squared norm · case 05
Squaring the total allows cancellation between samples.
Autocorrelation zero lag normalized · case 01
Sample count replaces zero-lag energy normalization.
Autocorrelation zero lag normalized · case 02
Sample count replaces zero-lag energy normalization.
Autocorrelation zero lag normalized · case 03
Sample count replaces zero-lag energy normalization.
Autocorrelation zero lag normalized · case 04
Sample count replaces zero-lag energy normalization.
Autocorrelation zero lag normalized · case 05
Sample count replaces zero-lag energy normalization.
Gregorian leap century rule · case 01
Divisibility by four ignores the century exception.
Gregorian leap century rule · case 02
Divisibility by four ignores the century exception.
Gregorian leap century rule · case 03
Divisibility by four ignores the century exception.
Gregorian leap century rule · case 04
Divisibility by four ignores the century exception.
Gregorian leap century rule · case 05
Divisibility by four ignores the century exception.
Calendar month length · case 01
A fixed thirty-day model loses long months and leap days.
Calendar month length · case 02
A fixed thirty-day model loses long months and leap days.
Calendar month length · case 03
A fixed thirty-day model loses long months and leap days.
Calendar month length · case 04
A fixed thirty-day model loses long months and leap days.
Calendar month length · case 05
A fixed thirty-day model loses long months and leap days.
Ordinal day of year · case 01
Month lengths are approximated by thirty days.
Ordinal day of year · case 02
Month lengths are approximated by thirty days.
Ordinal day of year · case 03
Month lengths are approximated by thirty days.
Ordinal day of year · case 04
Month lengths are approximated by thirty days.
Ordinal day of year · case 05
Month lengths are approximated by thirty days.
Days until next year exclusive · case 01
A fixed ordinary-year length also excludes the current day incorrectly.
Days until next year exclusive · case 02
A fixed ordinary-year length also excludes the current day incorrectly.
Days until next year exclusive · case 03
A fixed ordinary-year length also excludes the current day incorrectly.
Days until next year exclusive · case 04
A fixed ordinary-year length also excludes the current day incorrectly.
Days until next year exclusive · case 05
A fixed ordinary-year length also excludes the current day incorrectly.
Iso weekday monday one · case 01
A zero-based weekday is returned where Monday is one.
Iso weekday monday one · case 02
A zero-based weekday is returned where Monday is one.
Iso weekday monday one · case 03
A zero-based weekday is returned where Monday is one.
Iso weekday monday one · case 04
A zero-based weekday is returned where Monday is one.
Iso weekday monday one · case 05
A zero-based weekday is returned where Monday is one.
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 ↗