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
Min-sum offset can flip the message sign · case 01
Weak messages flip sign after the offset is applied.
Min-sum offset can flip the message sign · case 02
Weak messages flip sign after the offset is applied.
Min-sum offset can flip the message sign · case 03
Weak messages flip sign after the offset is applied.
Min-sum offset can flip the message sign · case 04
Weak messages flip sign after the offset is applied.
Min-sum offset can flip the message sign · case 05
Weak messages flip sign after the offset is applied.
Bit flipping stops before checking the last flip · case 01
Words fixed by the final permitted iteration are reported as failures.
Bit flipping stops before checking the last flip · case 02
Words fixed by the final permitted iteration are reported as failures.
Bit flipping stops before checking the last flip · case 03
Words fixed by the final permitted iteration are reported as failures.
Bit flipping stops before checking the last flip · case 04
Words fixed by the final permitted iteration are reported as failures.
Bit flipping stops before checking the last flip · case 05
Words fixed by the final permitted iteration are reported as failures.
Bit flipping flips only the first most-suspect bit · case 01
Decoding needs more iterations or fails where parallel flipping succeeds.
Bit flipping flips only the first most-suspect bit · case 02
Decoding needs more iterations or fails where parallel flipping succeeds.
Bit flipping flips only the first most-suspect bit · case 03
Decoding needs more iterations or fails where parallel flipping succeeds.
Bit flipping flips only the first most-suspect bit · case 04
Decoding needs more iterations or fails where parallel flipping succeeds.
Bit flipping flips only the first most-suspect bit · case 05
Decoding needs more iterations or fails where parallel flipping succeeds.
Code capability rounds the correction radius up · case 01
Even-distance codes are credited with one more correctable error.
Code capability rounds the correction radius up · case 02
Even-distance codes are credited with one more correctable error.
Code capability rounds the correction radius up · case 03
Even-distance codes are credited with one more correctable error.
Code capability rounds the correction radius up · case 04
Even-distance codes are credited with one more correctable error.
Code capability rounds the correction radius up · case 05
Even-distance codes are credited with one more correctable error.
Code capability weighs errors like erasures · case 01
Patterns with errors are declared decodable beyond the code capability.
Code capability weighs errors like erasures · case 02
Patterns with errors are declared decodable beyond the code capability.
Code capability weighs errors like erasures · case 03
Patterns with errors are declared decodable beyond the code capability.
Code capability weighs errors like erasures · case 04
Patterns with errors are declared decodable beyond the code capability.
Code capability weighs errors like erasures · case 05
Patterns with errors are declared decodable beyond the code capability.
Code capability measures distance between duplicate entries · case 01
A code list containing a repeated word reports distance 0.
Code capability measures distance between duplicate entries · case 02
A code list containing a repeated word reports distance 0.
Code capability measures distance between duplicate entries · case 03
A code list containing a repeated word reports distance 0.
Code capability measures distance between duplicate entries · case 04
A code list containing a repeated word reports distance 0.
Code capability measures distance between duplicate entries · case 05
A code list containing a repeated word reports distance 0.
Code capability claims d detectable errors · case 01
Detection capability is overstated by one.
Code capability claims d detectable errors · case 02
Detection capability is overstated by one.
Code capability claims d detectable errors · case 03
Detection capability is overstated by one.
Code capability claims d detectable errors · case 04
Detection capability is overstated by one.
Code capability claims d detectable errors · case 05
Detection capability is overstated by one.
Syndrome table keeps the heaviest coset member · case 01
Received words are "corrected" by high-weight patterns instead of the most likely error.
Syndrome table keeps the heaviest coset member · case 02
Received words are "corrected" by high-weight patterns instead of the most likely error.
Syndrome table keeps the heaviest coset member · case 03
Received words are "corrected" by high-weight patterns instead of the most likely error.
Syndrome table keeps the heaviest coset member · case 04
Received words are "corrected" by high-weight patterns instead of the most likely error.
Syndrome table keeps the heaviest coset member · case 05
Received words are "corrected" by high-weight patterns instead of the most likely error.
Syndrome table uses integer check sums · case 01
Syndromes of different cosets collide or miss the table.
Syndrome table uses integer check sums · case 02
Syndromes of different cosets collide or miss the table.
Syndrome table uses integer check sums · case 03
Syndromes of different cosets collide or miss the table.
Syndrome table uses integer check sums · case 04
Syndromes of different cosets collide or miss the table.
Syndrome table uses integer check sums · case 05
Syndromes of different cosets collide or miss the table.
Syndrome table ORs the error pattern into the word · case 01
Errors on bits that were received as 1 are never removed.
Syndrome table ORs the error pattern into the word · case 02
Errors on bits that were received as 1 are never removed.
Syndrome table ORs the error pattern into the word · case 03
Errors on bits that were received as 1 are never removed.
Syndrome table ORs the error pattern into the word · case 04
Errors on bits that were received as 1 are never removed.
Syndrome table ORs the error pattern into the word · case 05
Errors on bits that were received as 1 are never removed.
Depuncturing indexes the pattern by received count · case 01
Soft values are placed at shifted positions after the first punctured bit.
Depuncturing indexes the pattern by received count · case 02
Soft values are placed at shifted positions after the first punctured bit.
Depuncturing indexes the pattern by received count · case 03
Soft values are placed at shifted positions after the first punctured bit.
Depuncturing indexes the pattern by received count · case 04
Soft values are placed at shifted positions after the first punctured bit.
Depuncturing indexes the pattern by received count · case 05
Soft values are placed at shifted positions after the first punctured bit.
Depuncturing ignores the partial final pattern period · case 01
Blocks whose length is not a multiple of the pattern period are rejected.
Depuncturing ignores the partial final pattern period · case 02
Blocks whose length is not a multiple of the pattern period are rejected.
Depuncturing ignores the partial final pattern period · case 03
Blocks whose length is not a multiple of the pattern period are rejected.
Depuncturing ignores the partial final pattern period · case 04
Blocks whose length is not a multiple of the pattern period are rejected.
Depuncturing ignores the partial final pattern period · case 05
Blocks whose length is not a multiple of the pattern period are rejected.
QR format encoder omits the 0x5412 mask · case 01
Format words disagree with printed symbols; decoding finds nothing within distance 3.
QR format encoder omits the 0x5412 mask · case 02
Format words disagree with printed symbols; decoding finds nothing within distance 3.
QR format encoder omits the 0x5412 mask · case 03
Format words disagree with printed symbols; decoding finds nothing within distance 3.
QR format encoder omits the 0x5412 mask · case 04
Format words disagree with printed symbols; decoding finds nothing within distance 3.
QR format encoder omits the 0x5412 mask · case 05
Format words disagree with printed symbols; decoding finds nothing within distance 3.
QR format encoder numbers error-correction levels in order · case 01
Every level except H decodes as a different level.
QR format encoder numbers error-correction levels in order · case 02
Every level except H decodes as a different level.
QR format encoder numbers error-correction levels in order · case 03
Every level except H decodes as a different level.
QR format encoder numbers error-correction levels in order · case 04
Every level except H decodes as a different level.
QR format encoder numbers error-correction levels in order · case 05
Every level except H decodes as a different level.
QR format BCH division skips the lowest step · case 01
Remainders are left unreduced for some data values.
QR format BCH division skips the lowest step · case 02
Remainders are left unreduced for some data values.
QR format BCH division skips the lowest step · case 03
Remainders are left unreduced for some data values.
QR format BCH division skips the lowest step · case 04
Remainders are left unreduced for some data values.
QR format BCH division skips the lowest step · case 05
Remainders are left unreduced for some data values.
QR format decoder only accepts two bit errors · case 01
Format words with three bit errors are rejected although the code corrects them.
QR format decoder only accepts two bit errors · case 02
Format words with three bit errors are rejected although the code corrects them.
QR format decoder only accepts two bit errors · case 03
Format words with three bit errors are rejected although the code corrects them.
QR format decoder only accepts two bit errors · case 04
Format words with three bit errors are rejected although the code corrects them.
QR format decoder only accepts two bit errors · case 05
Format words with three bit errors are rejected although the code corrects them.
Row retirement never expires old corrected errors · case 01
Rows with sparse, harmless corrected errors are retired.
Row retirement never expires old corrected errors · case 02
Rows with sparse, harmless corrected errors are retired.
Row retirement never expires old corrected errors · case 03
Rows with sparse, harmless corrected errors are retired.
Row retirement never expires old corrected errors · case 04
Rows with sparse, harmless corrected errors are retired.
Row retirement never expires old corrected errors · case 05
Rows with sparse, harmless corrected errors are retired.
Row retirement pools corrected errors across rows · case 01
Unrelated rows are retired because their neighbours had errors.
Row retirement pools corrected errors across rows · case 02
Unrelated rows are retired because their neighbours had errors.
Row retirement pools corrected errors across rows · case 03
Unrelated rows are retired because their neighbours had errors.
Row retirement pools corrected errors across rows · case 04
Unrelated rows are retired because their neighbours had errors.
Row retirement pools corrected errors across rows · case 05
Unrelated rows are retired because their neighbours had errors.
Row retirement re-retires rows on later events · case 01
Retirement timestamps move later as more errors arrive on an already retired row.
Row retirement re-retires rows on later events · case 02
Retirement timestamps move later as more errors arrive on an already retired row.
Row retirement re-retires rows on later events · case 03
Retirement timestamps move later as more errors arrive on an already retired row.
Row retirement re-retires rows on later events · case 04
Retirement timestamps move later as more errors arrive on an already retired row.
Row retirement re-retires rows on later events · case 05
Retirement timestamps move later as more errors arrive on an already retired row.
Row retirement counts uncorrectable errors like corrected ones · case 01
Rows with an uncorrectable error stay in service until the threshold is reached.
Row retirement counts uncorrectable errors like corrected ones · case 02
Rows with an uncorrectable error stay in service until the threshold is reached.
Row retirement counts uncorrectable errors like corrected ones · case 03
Rows with an uncorrectable error stay in service until the threshold is reached.
Row retirement counts uncorrectable errors like corrected ones · case 04
Rows with an uncorrectable error stay in service until the threshold is reached.
Row retirement counts uncorrectable errors like corrected ones · case 05
Rows with an uncorrectable error stay in service until the threshold is reached.
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 ↗