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
Pcm16 saturation before encoding · case 01
Overflow wraps a large positive sample into a negative amplitude.
Pcm16 saturation before encoding · case 02
Overflow wraps a large positive sample into a negative amplitude.
Pcm16 saturation before encoding · case 03
Overflow wraps a large positive sample into a negative amplitude.
Pcm16 saturation before encoding · case 04
Overflow wraps a large positive sample into a negative amplitude.
Pcm16 saturation before encoding · case 05
Overflow wraps a large positive sample into a negative amplitude.
Float to pcm16 endpoint clamp · case 01
Positive full scale quantizes to unrepresentable 32768 without final clamping.
Float to pcm16 endpoint clamp · case 02
Positive full scale quantizes to unrepresentable 32768 without final clamping.
Float to pcm16 endpoint clamp · case 03
Positive full scale quantizes to unrepresentable 32768 without final clamping.
Float to pcm16 endpoint clamp · case 04
Positive full scale quantizes to unrepresentable 32768 without final clamping.
Float to pcm16 endpoint clamp · case 05
Positive full scale quantizes to unrepresentable 32768 without final clamping.
Pcm16 to float asymmetric range · case 01
Normalization by positive maximum pushes negative full scale below -1.
Pcm16 to float asymmetric range · case 02
Normalization by positive maximum pushes negative full scale below -1.
Pcm16 to float asymmetric range · case 03
Normalization by positive maximum pushes negative full scale below -1.
Pcm16 to float asymmetric range · case 04
Normalization by positive maximum pushes negative full scale below -1.
Pcm16 to float asymmetric range · case 05
Normalization by positive maximum pushes negative full scale below -1.
Unsigned eight bit silence padding · case 01
Zero bytes pad unsigned audio with maximum negative amplitude.
Unsigned eight bit silence padding · case 02
Zero bytes pad unsigned audio with maximum negative amplitude.
Unsigned eight bit silence padding · case 03
Zero bytes pad unsigned audio with maximum negative amplitude.
Unsigned eight bit silence padding · case 04
Zero bytes pad unsigned audio with maximum negative amplitude.
Unsigned eight bit silence padding · case 05
Zero bytes pad unsigned audio with maximum negative amplitude.
Interleave stereo frame order · case 01
Concatenating channel planes does not produce interleaved frames.
Interleave stereo frame order · case 02
Concatenating channel planes does not produce interleaved frames.
Interleave stereo frame order · case 03
Concatenating channel planes does not produce interleaved frames.
Interleave stereo frame order · case 04
Concatenating channel planes does not produce interleaved frames.
Interleave stereo frame order · case 05
Concatenating channel planes does not produce interleaved frames.
Deinterleave stereo stride · case 01
Halving the array treats interleaved input as planar.
Deinterleave stereo stride · case 02
Halving the array treats interleaved input as planar.
Deinterleave stereo stride · case 03
Halving the array treats interleaved input as planar.
Deinterleave stereo stride · case 04
Halving the array treats interleaved input as planar.
Deinterleave stereo stride · case 05
Halving the array treats interleaved input as planar.
Sample count to frame count · case 01
Sample count is reported as frame count in multichannel audio.
Sample count to frame count · case 02
Sample count is reported as frame count in multichannel audio.
Sample count to frame count · case 03
Sample count is reported as frame count in multichannel audio.
Sample count to frame count · case 04
Sample count is reported as frame count in multichannel audio.
Sample count to frame count · case 05
Sample count is reported as frame count in multichannel audio.
Frame slice multiplies channel stride · case 01
Frame offsets and lengths are applied as individual sample offsets.
Frame slice multiplies channel stride · case 02
Frame offsets and lengths are applied as individual sample offsets.
Frame slice multiplies channel stride · case 03
Frame offsets and lengths are applied as individual sample offsets.
Frame slice multiplies channel stride · case 04
Frame offsets and lengths are applied as individual sample offsets.
Frame slice multiplies channel stride · case 05
Frame offsets and lengths are applied as individual sample offsets.
Retain partial frame carry · case 01
Chunk processing discards the previous incomplete frame and current trailing samples.
Retain partial frame carry · case 02
Chunk processing discards the previous incomplete frame and current trailing samples.
Retain partial frame carry · case 03
Chunk processing discards the previous incomplete frame and current trailing samples.
Retain partial frame carry · case 04
Chunk processing discards the previous incomplete frame and current trailing samples.
Retain partial frame carry · case 05
Chunk processing discards the previous incomplete frame and current trailing samples.
Reverse audio by frames · case 01
Reversing scalar samples also reverses channel order inside each frame.
Reverse audio by frames · case 02
Reversing scalar samples also reverses channel order inside each frame.
Reverse audio by frames · case 03
Reversing scalar samples also reverses channel order inside each frame.
Reverse audio by frames · case 04
Reversing scalar samples also reverses channel order inside each frame.
Reverse audio by frames · case 05
Reversing scalar samples also reverses channel order inside each frame.
Channel remap framewise · case 01
A configured channel-routing map is ignored.
Channel remap framewise · case 02
A configured channel-routing map is ignored.
Channel remap framewise · case 03
A configured channel-routing map is ignored.
Channel remap framewise · case 04
A configured channel-routing map is ignored.
Channel remap framewise · case 05
A configured channel-routing map is ignored.
Replace one channel · case 01
A mono channel update overwrites every channel in its frame.
Replace one channel · case 02
A mono channel update overwrites every channel in its frame.
Replace one channel · case 03
A mono channel update overwrites every channel in its frame.
Replace one channel · case 04
A mono channel update overwrites every channel in its frame.
Replace one channel · case 05
A mono channel update overwrites every channel in its frame.
Mono to multichannel duplication · case 01
Duplicating the whole waveform produces channel-major storage.
Mono to multichannel duplication · case 02
Duplicating the whole waveform produces channel-major storage.
Mono to multichannel duplication · case 03
Duplicating the whole waveform produces channel-major storage.
Mono to multichannel duplication · case 04
Duplicating the whole waveform produces channel-major storage.
Mono to multichannel duplication · case 05
Duplicating the whole waveform produces channel-major storage.
Frame padding preserves shape · case 01
Padding inserts scalar zero where downstream code expects a complete frame.
Frame padding preserves shape · case 02
Padding inserts scalar zero where downstream code expects a complete frame.
Frame padding preserves shape · case 03
Padding inserts scalar zero where downstream code expects a complete frame.
Frame padding preserves shape · case 04
Padding inserts scalar zero where downstream code expects a complete frame.
Frame padding preserves shape · case 05
Padding inserts scalar zero where downstream code expects a complete frame.
Read cursor advances consumed · case 01
Read cursor advances beyond end by requested rather than consumed count.
Read cursor advances consumed · case 02
Read cursor advances beyond end by requested rather than consumed count.
Read cursor advances consumed · case 03
Read cursor advances beyond end by requested rather than consumed count.
Read cursor advances consumed · case 04
Read cursor advances beyond end by requested rather than consumed count.
Read cursor advances consumed · case 05
Read cursor advances beyond end by requested rather than consumed count.
Underrun zero fill · case 01
A device callback returns fewer samples than the hardware requested.
Underrun zero fill · case 02
A device callback returns fewer samples than the hardware requested.
Underrun zero fill · case 03
A device callback returns fewer samples than the hardware requested.
Underrun zero fill · case 04
A device callback returns fewer samples than the hardware requested.
Underrun zero fill · case 05
A device callback returns fewer samples than the hardware requested.
Overrun drops oldest frames · case 01
Truncating the tail retains stale buffered audio and drops newest input.
Overrun drops oldest frames · case 02
Truncating the tail retains stale buffered audio and drops newest input.
Overrun drops oldest frames · case 03
Truncating the tail retains stale buffered audio and drops newest input.
Overrun drops oldest frames · case 04
Truncating the tail retains stale buffered audio and drops newest input.
Overrun drops oldest frames · case 05
Truncating the tail retains stale buffered audio and drops newest input.
Stereo downmix average · case 01
Adding two full-scale channels doubles expected mono loudness.
Stereo downmix average · case 02
Adding two full-scale channels doubles expected mono loudness.
Stereo downmix average · case 03
Adding two full-scale channels doubles expected mono loudness.
Stereo downmix average · case 04
Adding two full-scale channels doubles expected mono loudness.
Stereo downmix average · case 05
Adding two full-scale channels doubles expected mono loudness.
Mix pad shorter source · case 01
Zip truncation loses the longer source tail.
Mix pad shorter source · case 02
Zip truncation loses the longer source tail.
Mix pad shorter source · case 03
Zip truncation loses the longer source tail.
Mix pad shorter source · case 04
Zip truncation loses the longer source tail.
Mix pad shorter source · case 05
Zip truncation loses the longer source tail.
Mix saturate after sum · case 01
Summed audio exceeds normalized output range.
Mix saturate after sum · case 02
Summed audio exceeds normalized output range.
Mix saturate after sum · case 03
Summed audio exceeds normalized output range.
Mix saturate after sum · case 04
Summed audio exceeds normalized output range.
Mix saturate after sum · case 05
Summed audio exceeds normalized output range.
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 ↗