Sandbox::https://github.com/questdb/questdb/issues

SAMPLE BY FILL with a non-integer literal on a LONG aggregate throws UnsupportedOperationException at runtime

questdb/questdb#7867 · opened

on Oct 11, 2026

INNER join ON condition that reads no column drops the unmatched rows of a later RIGHT or FULL join

questdb/questdb#7866 · opened

on Oct 11, 2026

extract(epoch) is off by one second for pre-1970 timestamps with a fractional part

questdb/questdb#7865 · opened

on Oct 10, 2026

Simple CASE with a bracketed, function-call, cast or CASE operand runs as a searched CASE and returns wrong rows

questdb/questdb#7864 · opened

on Oct 10, 2026

HORIZON JOIN resolves an ON key on an earlier HORIZON JOIN's table to the left table's column

questdb/questdb#7857 · opened

on Oct 10, 2026

LATEST ON over an INNER join returns a row that is not the latest when two ON keys share a column

questdb/questdb#7856 · opened

on Oct 9, 2026

ORDER BY timestamp DESC over an INNER join followed by an ASOF or LT JOIN fails with "left side of time series join doesn't have ASC timestamp order"

questdb/questdb#7854 · opened

on Oct 9, 2026

ASOF JOIN and SAMPLE BY over a sub-query fail with "TIMESTAMP column is required" when the optimiser runs the sub-query's FROM table last

questdb/questdb#7853 · opened

on Oct 9, 2026

Live view that emits no rows keeps its base table's WAL from being purged

questdb/questdb#7852 · opened

on Oct 9, 2026

Live view double-counts a base commit and fails reads after DEDUP ENABLE on its base table

questdb/questdb#7851 · opened

on Oct 9, 2026

INSERT of a FLOAT just above the INT range stores 2147483647 instead of failing; FLOAT and DOUBLE into LONG the same

questdb/questdb#7849 · opened

on Oct 9, 2026

1

An ILP over UDP line that adds a column to a WAL table leaves the table suspended at the next SQL insert

questdb/questdb#7848 · opened

on Oct 9, 2026

ILP over UDP silently loses rows when two tables are followed by the same third table in one commit cycle

questdb/questdb#7847 · opened

on Oct 9, 2026

With cairo.commit.mode=sync, WAL ingest flushes every filled 1 MB column page synchronously and runs 10-25 times slower than nosync

questdb/questdb#7846 · opened

on Oct 9, 2026

SYNC commit mode: ALTER COLUMN ADD INDEX does not flush the new index files, so a power loss can break queries on the indexed column

questdb/questdb#7845 · opened

on Oct 9, 2026

SYNC commit mode: ALTER COLUMN TYPE to SYMBOL does not flush the new symbol map or index files, so a power loss can make the table unreadable

questdb/questdb#7844 · opened

on Oct 9, 2026

SYNC commit mode: ALTER COLUMN TYPE does not flush the converted column files, so a power loss can lose the column

questdb/questdb#7843 · opened

on Oct 9, 2026

After ALTER TABLE ... REBASE WAL, the table's min timestamp can become 1970-01-01

questdb/questdb#7842 · opened

on Oct 9, 2026

A crash during an out-of-order commit leaves old POSTING index files on disk for good

questdb/questdb#7841 · opened

on Oct 9, 2026

A crash during ATTACH PARTITION deletes the partition being attached

questdb/questdb#7840 · opened

on Oct 9, 2026

SAMPLE BY FILL fails or returns wrong rows when an aggregate is aliased to the designated timestamp name

questdb/questdb#7838 · opened

on Oct 9, 2026

1

Filter on a SYMBOL value returns no rows while a writer holds uncommitted rows with new symbols

questdb/questdb#7837 · opened

on Oct 9, 2026

LATERAL sub-query over a join fails with "Ambiguous column" when its tables share a column name and WHERE has two correlated conditions

questdb/questdb#7836 · opened

on Oct 9, 2026

LATERAL sub-query that fails to compile leaves the source of its shared cursor allocated

questdb/questdb#7835 · opened

on Oct 9, 2026

SAMPLE BY puts a NULL designated timestamp that isn't in the first row into the wrong bucket on the SAMPLE BY cursor path

questdb/questdb#7834 · opened

on Oct 8, 2026

Navigation Menu

Repository navigation

Issues

Issues

SAMPLE BY FILL with a non-integer literal on a LONG aggregate throws UnsupportedOperationException at runtime

INNER join ON condition that reads no column drops the unmatched rows of a later RIGHT or FULL join

extract(epoch) is off by one second for pre-1970 timestamps with a fractional part

Simple CASE with a bracketed, function-call, cast or CASE operand runs as a searched CASE and returns wrong rows

HORIZON JOIN resolves an ON key on an earlier HORIZON JOIN's table to the left table's column

LATEST ON over an INNER join returns a row that is not the latest when two ON keys share a column

ORDER BY timestamp DESC over an INNER join followed by an ASOF or LT JOIN fails with "left side of time series join doesn't have ASC timestamp order"

ASOF JOIN and SAMPLE BY over a sub-query fail with "TIMESTAMP column is required" when the optimiser runs the sub-query's FROM table last

Live view that emits no rows keeps its base table's WAL from being purged

Live view double-counts a base commit and fails reads after DEDUP ENABLE on its base table

INSERT of a FLOAT just above the INT range stores 2147483647 instead of failing; FLOAT and DOUBLE into LONG the same

An ILP over UDP line that adds a column to a WAL table leaves the table suspended at the next SQL insert

ILP over UDP silently loses rows when two tables are followed by the same third table in one commit cycle

With cairo.commit.mode=sync, WAL ingest flushes every filled 1 MB column page synchronously and runs 10-25 times slower than nosync

SYNC commit mode: ALTER COLUMN ADD INDEX does not flush the new index files, so a power loss can break queries on the indexed column

SYNC commit mode: ALTER COLUMN TYPE to SYMBOL does not flush the new symbol map or index files, so a power loss can make the table unreadable

SYNC commit mode: ALTER COLUMN TYPE does not flush the converted column files, so a power loss can lose the column

After ALTER TABLE ... REBASE WAL, the table's min timestamp can become 1970-01-01

A crash during an out-of-order commit leaves old POSTING index files on disk for good

A crash during ATTACH PARTITION deletes the partition being attached

SAMPLE BY FILL fails or returns wrong rows when an aggregate is aliased to the designated timestamp name

Filter on a SYMBOL value returns no rows while a writer holds uncommitted rows with new symbols

LATERAL sub-query over a join fails with "Ambiguous column" when its tables share a column name and WHERE has two correlated conditions

LATERAL sub-query that fails to compile leaves the source of its shared cursor allocated

SAMPLE BY puts a NULL designated timestamp that isn't in the first row into the wrong bucket on the SAMPLE BY cursor path