microsoft/fhir-server release/4.0.766

4.0.766

Repository: microsoft/fhir-server

Tag: release/4.0.766

Published: 2026-05-18T01:09:08Z

Prerelease: no

Release notes:

What's Changed

Azure Health Data Services (SQL)

• [Fix null cast in QueryPlanReuseChecker #5570](https://github.com/microsoft/fhir-server/pull/5570)

Fixed a null reference exception in QueryPlanReuseChecker when reading the skew value from the database by safely handling DBNull values. Also adds an early exit when the feature is disabled.

• [Return 404 for mis-cased FHIR resource type segments #5530](https://github.com/microsoft/fhir-server/pull/5530)

Requests with incorrectly-cased resource types (e.g., /patient instead of /Patient) now return 404 Not Found instead of 405 Method Not Allowed.

Common to both Azure API for FHIR and Azure Health Data Services

• [Add tests to cover birthdate search semantics #5561](https://github.com/microsoft/fhir-server/pull/5561)

Added E2E tests to document and preserve the current behavior of patient birthdate searches using partial dates (year, year-month, year-month-day). These tests capture an existing deviation from the FHIR R4 equality containment rule, laying groundwork for a future conformance fix.

• [Upsert profiles on ValidateFixture #5568](https://github.com/microsoft/fhir-server/pull/5568)

Fixed a test fixture issue where StructureDefinition, ValueSet, and CodeSystem profile resources were being created anew on every test run, causing accumulation in persistent environments. The fixture now uses upsert (PUT) with stable IDs instead.

• [Ensure Unique Query Parameters #5557](https://github.com/microsoft/fhir-server/pull/5557)

Deduplicated exact (name, value) query parameter pairs before expression parsing, preventing redundant SQL CTEs from being generated when the same search parameter is repeated in a request. This is semantically safe because repeated AND predicates are idempotent per the FHIR specification.

Other Changes

• [Reverted unnecessary reindex test changes #5563](https://github.com/microsoft/fhir-server/pull/5563)

Reverted recent CI-driven changes to reindex E2E tests and restored more reliable test setup by adding IAsyncLifetime initialization to clean up leftover resources before each run. Also adjusted CI pipeline settings (polling frequency, cache refresh intervals, replica count) to reduce flakiness.

• [Increase SQL CPUs in PR pipeline #5575](https://github.com/microsoft/fhir-server/pull/5575)

Upgraded the SQL database SKU used in the PR pipeline from GP_Gen5_4 to GP_Gen5_8 to improve pipeline performance and reduce test timeouts under load.

Full Changelog: https://github.com/microsoft/fhir-server/compare/release/4.0.759...release/4.0.766