Stable 26.8 - Bump to latest - #2476

…3829

Backport ClickHouse#123829 to 26.8: Do not read a JOIN input that cannot contribute under a WHERE no row passes

…filters below `LIMIT BY`

…es evaluate a predicate with OR

Release v26.8.20.9-lts

… a subquery that does not see it when updating

…ributed table over a Distributed table

…9630

Backport ClickHouse#119630 to 26.8: Do not expand a WITH element into a subquery that does not see it when updating

…es before casting to an engine type

…rom peer-declared sizes when reading settings from the Native protocol""

…1332

Backport ClickHouse#121332 to 26.8: Resolve lazily loaded table proxies before casting to an engine type

…0931

Backport ClickHouse#120931 to 26.8: Revert "Revert "Do not allocate from peer-declared sizes when reading settings from the Native protocol""

…1675

Backport ClickHouse#121675 to 26.8: Add a setting to control pushing filters below `LIMIT BY`

…ge inserts from a different dictionary

…3900

Backport ClickHouse#123900 to 26.8: Faster short `LowCardinality` range inserts from a different dictionary

…-hint-26.8

26.8: Fix `DISTINCT` output chunks reserving the memory of their whole input chunk

… that drops `Nullable`

…Exact on this branch

tryConvertFieldToTypeExact comes from ClickHouse#121078, which is not on this branch.
At this call site only a widening conversion (DateTime or lower-scale
DateTime64 to DateTime64) reaches it, which is exact or overflows to Null,
so the strict tryConvertFieldToType gives the same result.

In 05330, the "implied condition is not duplicated" case compares `a` with a
DateTime64 constant instead of a DateTime one: on this branch the existing
`a < <DateTime>` conjunct hits TYPE_MISMATCH in tryConvertToColumnType
(ClickHouse#117903, fixed on master by ClickHouse#121078), so it is not seen as implying the
derived bound.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…reading `groupBitmap` state

…4220

Backport ClickHouse#124220 to 26.8: Use skip indexes through a `CAST` that drops `Nullable`

…3514

Backport ClickHouse#123514 to 26.8: Fix mv source engine checks

…6120

Backport ClickHouse#126120 to 26.8: Fix out-of-bounds accesses in the replxx line editor

…sts after ExpiredToken

… codec chain

…d in `05316_named_collection_dictionary_override_access`

…3195

Backport ClickHouse#123195 to 26.8: Retry S3 backup and restore requests after ExpiredToken

…3877

Backport ClickHouse#113877 to 26.8: Fix out-of-bounds write on a long codec chain

…ivilege check

…4075

Backport ClickHouse#124075 to 26.8: Leave only loadable objects behind in `05316_named_collection_dictionary_override_access`

…n estimating subcolumn sizes for PREWHERE

…RT when a row policy of a table read by a materialized view uses `IN <source table>`

…8866

Backport ClickHouse#118866 to 26.8: Require SELECT on the source when CREATE TABLE AS inherits credentials

…DataSourceDescription` in every backup entry

…6046

Backport ClickHouse#126046 to 26.8: Do not read part dictionaries when estimating subcolumn sizes for PREWHERE

…5893

Backport ClickHouse#125893 to 26.8: Fix a crash and lost rows on INSERT when a row policy of a table read by a materialized view uses `IN <source table>`

…6388

Backport ClickHouse#126388 to 26.8: Don't keep a copy of the disk's `DataSourceDescription` in every backup entry

…n Protobuf schemas

…Redis replies

…t on `ENUM` without values

…ant/Dynamic/JSON columns containing only NULLs

…6550

Backport ClickHouse#126550 to 26.8: Limit nesting depth of messages in Protobuf schemas

…6547

Backport ClickHouse#126547 to 26.8: Limit nesting depth of arrays in Redis replies

…0776

Backport ClickHouse#120776 to 26.8: Fix uninitialized offsets in Variant/Dynamic/JSON columns containing only NULLs

…ion after a failed drain into the shared table

…6159

Backport ClickHouse#126159 to 26.8: Fix segfault in `MySQLDump` format on `ENUM` without values

`ASTSelectQuery::limit_after_all` does not exist in 26.8, and `splitAndModifyMutationCommands`
takes `suitable_for_ttl_optimization` here. 26.8 ignores inline setting history in `DECLARE`,
so the new settings are recorded in the 26.8 block of `SettingsChangesHistory.cpp`.

ClickHouse#124274

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…6126

Backport ClickHouse#126126 to 26.8: Fix segfault in adaptive aggregation after a failed drain into the shared table

…2718

Backport ClickHouse#122718 to 26.8: Coalesce adjacent singleton row-mask updates during part mutation

Release v26.8.32.14-lts