Added support for local distributed tables in join queries

[glookka]

Added support for local distributed tables in join queries

[Copilot]

Pull Request Overview

Added support for performing joins over local partitions of distributed tables by propagating the original “parent” table name through queue creation, join sorter, and search handler logic.

• Extended queue creator and multi-queue interfaces with a szParent parameter to carry the parent index name.
• Updated join sorter and filter-checking logic to use the passed parent name for distributed-local joins.
• Modified search handler to build and track JoinedIndexes_t with parent metadata.
• Added an SQL test in test/test_278/test.xml for inner joins over a local distributed table.

Reviewed Changes

Copilot reviewed 7 out of 8 changed files in this pull request and generated 2 comments.

File	Description
test/test_278/test.xml	Added inner-join test case for a local-distributed table d.
src/queuecreator.h/.cpp	Added const char* szParent to queue-creation APIs and constructors.
src/joinsorter.h/.cpp	Propagated parent table names into join sorter constructors and filters.
src/daemon/search_handler.h/.cpp	Introduced JoinedIndexes_t, updated joined index population, and search flow.

Comments suppressed due to low confidence (3)

src/queuecreator.h:53

• Adding the szParent parameter changes the public API signature without a default argument or overload, breaking existing callers. Please add a default value (= nullptr) or provide an overload to maintain backward compatibility.

void			sphCreateMultiQueue ( const SphQueueSettings_t & tQueue, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> & dSorters, VecTraits_T<CSphString> & dErrors, SphQueueRes_t & tRes, StrVec_t * pExtra, QueryProfile_c * pProfile, const char * szParent );

test/test_278/test.xml:665

• [nitpick] The new test covers only an inner join on the distributed-local table. Consider adding tests for left joins and failure scenarios (e.g., joining on a distributed table with remote agents) to ensure full coverage of the new feature.

<!-- join over distributed left table -->

src/joinsorter.cpp:722

• [nitpick] GetJoinedIndexName() allocates a new CSphString on every call, which may be expensive in hot paths. Consider returning a const CSphString& or caching the result to reduce heap allocations.

	CSphString	GetJoinedIndexName() const	{ return m_sRightIndexSchemaName.IsEmpty() ?  m_pJoinedIndex->GetName() : m_sRightIndexSchemaName; }

[Copilot]

Storing raw const char* pointers to external CSphString data can lead to dangling pointers when the original string is modified or destroyed. Consider using a CSphString member for m_szParent to own the string and ensure safety.

Suggested change

	const char * m_szParent = nullptr;
	CSphString m_sParent;

[Copilot]

The error message literal opens with a single quote but does not clearly close it, leading to mismatched quotes. Update the literal or add a trailing apostrophe in the string to ensure the output is consistent.

[github-actions]

Windows test results

    3 files      3 suites   24m 46s ⏱️
1 083 tests 1 031 ✅ 52 💤 0 ❌
1 087 runs  1 035 ✅ 52 💤 0 ❌

Results for commit ab59480.

♻️ This comment has been updated with latest results.