The Queries tabs now explain how to turn on query statistics — When query statistics are unavailable, the Slow Log and Stats tabs show the missing setup step, copy-paste commands, and a link to the docs. The hint distinguishes a missing extension from a database library that still needs to be loaded.
Optional auto-enable of query statistics for development and staging — Set database.auto-create-query-stats-extension = true to let IdentityScribe create the query statistics extension at startup after PostgreSQL has loaded the required library. The default is auto (on in development, off everywhere else), so production is unchanged unless you opt in. See Enabling query statistics.
Changed
Query diagnostic reports explain multi-field substring searches more clearly — Support reports now show whether each searched field used the faster path or a safe fallback, with the reason when fallback was required. These rows previously reported that estimation was not captured, which made shared reports harder to act on.
Fixed
Licensed deployments start without configuring an audience — License verification no longer fails when license.audience is left unset. The licensee your license was issued to is read from the signed license itself, so a valid license works out of the box. Set license.audience (or the SCRIBE_LICENSE_AUDIENCE environment variable) only when you want a deployment to match a specific licensee name. Previously, a deployment that did not set license.audience could fail to start with an Invalid audience error.
Prefix searches across substring-search fields are faster — Searches that combine several “starts with” conditions across configured search fields no longer fall back to full-directory work when a faster bounded lookup is safe. Large-directory searches return sooner with the same matches, ordering, and cursor pagination as before. Broad searches keep the safer path automatically, and no configuration change is required.
Query statistics appear as soon as the extension is enabled — Turning on query statistics after IdentityScribe was first deployed no longer needs a redeploy or a follow-up migration; the Queries tabs and /observe/stats/queries pick it up on the next refresh. Creating the extension before loading the database library no longer makes the endpoint fail; it reports the missing setup step and points to the fix.
Long prefix searches no longer admit neighboring values — Prefix searches with very long values now check the full requested value after using accelerated narrowing, so entries that only share the stored leading segment are no longer included. Normal short prefix searches keep the same fast path.
GraphQL field-name docs match the served schema for mixed-case hyphenated attributes — The GraphQL guide now shows the exact field name exposed for attributes such as DirXML-EntitlementRef, so generated examples, schema exploration, and documentation agree.
