Skip to content

Azure.Search.Documents_12.1.0-beta.2

Pre-release
Pre-release

Choose a tag to compare

View all tags
@azure-sdk-automation azure-sdk-automation released this 28 Aug 23:04
Azure.Search.Documents_12.1.0-beta.2
9e93a8e
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

12.1.0-beta.2 (2026-08-27)

Features Added

  • Added support for the 2026-08-01-preview service version (now the default/latest version).
  • Added typed server-sent event streaming for knowledge base retrieval through KnowledgeBaseRetrievalClient.RetrieveStreamAsync, including typed retrieval, activity, answer, reference, completion, error, and unknown event models.
  • Added ListingSearchType along with search, pageSize, and searchType parameters on the resource listing methods (GetIndexes, GetAliases, GetKnowledgeBases, GetKnowledgeSources, GetSynonymMaps, GetIndexers, GetDataSourceConnections, GetSkillsets, and GetIndexStatsSummary).
  • Added Microsoft Entra application authentication for Work IQ knowledge sources via WorkIQKnowledgeSourceParameters and EntraAppAuthentication.
  • Added SearchIndexKnowledgeSource query-hint and boost configuration types: SearchIndexKnowledgeSourceQueryHints, SearchIndexKnowledgeSourceFilterHint, SearchIndexKnowledgeSourceFieldValueBoost, and SearchIndexKnowledgeSourceMultiWordExpressionBoost.
  • Added KnowledgeSourceResultsProcessing for controlling results processing on knowledge sources, retrieval requests, and McpServerTool, and added NeverQuerySource for request-local source exclusion.
  • Added KnowledgeBaseRetrieveDefaults, KnowledgeBaseQueryHintProcessing, KnowledgeRetrievalAutoReasoningEffort, and ServedImage for knowledge base retrieval.
  • Added KnowledgeBaseActivityRecordModel, exposing the model name and deployment id for model activity records, along with activity start/completion timestamps and logical reasoning effort details.
  • Added citation URLs and served-image metadata to supported knowledge base references, and added SearchSensitivityLabelInfo to Work IQ references.
  • Added knowledge source file multipart upload and update operations, with FileUploadMetadata, UploadKnowledgeSourceFileMultipartRequest, and UpdateKnowledgeSourceFileRequest, plus file prefix, metadata, parsing mode, and extraction mode properties.
  • Added KnowledgeSourceNetworkAccessMode and FileKnowledgeSourceExtractionMode.
  • Added tags and retrieval defaults to KnowledgeBase, MaxVectorIndexSizePerIndexInBytes to SearchServiceLimits, and Gpt56Sol, Gpt56Terra, and Gpt56Luna to AzureOpenAIModelName.
  • Added queryWorkIQSourceAuthorization to knowledge base retrieval methods for Work IQ on-behalf-of authentication.

Breaking Changes

These changes are relative to the previous preview release, 12.1.0-beta.1.

  • Removed SearchClientOptions.ServiceVersion.V2026_05_01_Preview; use V2026_08_01_Preview instead.
  • The resource listing methods (GetIndexes, GetAliases, GetKnowledgeBases, GetKnowledgeSources, GetSynonymMaps, GetIndexers, GetDataSourceConnections, GetSkillsets, and GetIndexStatsSummary) replaced the top, skip, and count parameters with search, pageSize, and searchType.
  • WorkIQKnowledgeSource now requires a WorkIQKnowledgeSourceParameters argument (Entra application authentication) in its constructor.
  • The model activity records (KnowledgeBaseModelQueryPlanningActivityRecord, KnowledgeBaseModelAnswerSynthesisActivityRecord, and KnowledgeBaseModelWebSummarizationActivityRecord) now expose model information through the Model property (KnowledgeBaseActivityRecordModel) instead of the ModelName string.
  • Removed McpServerToolInclusionMode and the McpServerTool.InclusionMode property; MCP tool behavior is now configured through ResultsProcessing and OutputParsing.
  • Removed WorkIQAttribution and the KnowledgeBaseWorkIQReference.Attributions property; use SearchSensitivityLabelInfo for the current Work IQ reference metadata.

Bugs Fixed

  • Fixed System.Text.Json.JsonException / InvalidOperationException: Cannot skip tokens on partial JSON thrown when asynchronously deserializing a SearchDocument (for example SearchAsync<SearchDocument>) whose result elements are large and contain OData (@search.*) properties. The internal converter now uses Utf8JsonReader.TrySkip() instead of Skip() so these properties can be skipped even when the converter is invoked over a non-final JSON block. (#40768)
Assets 2
Loading