Skip to content

Quine Release 2.2.1

Latest
Latest

Choose a tag to compare

View all tags
@listba listba released this 28 Oct 14:55
· 13 commits to main since this release
v2.2.1
0ab07cf

Quine 2.2.1

Enhancements:

  • Quine now integrates large language models into its data streams, so AI can interpret unstructured text and make decisions on live data, with the results applied to the graph in real time and no separate AI pipeline to build or maintain:
    • llm.prompt turns free text into structured graph data, and llm.decide delivers probability-scored judgments for classification and triage at streaming speed.
    • Supports OpenAI-compatible services, including self-hosted models, as well as Anthropic and Jev. Credentials stay in configuration, out of queries.
    • Calls can fail over to another configured model, and ingest and standing queries retry through model outages. Usage limits help control cost.
  • Investigating what the graph looked like at a past moment is now point-and-click. A new Query time dialog takes natural dates or relative times, and an on-screen reminder shows whenever you are viewing the past.
  • Watching standing query results appear live on the graph is now easy and fast to set up: graph feeds suggest their own query and name, and can start drawing the moment they are saved.
  • Queries that call purgeNode must switch to DETACH DELETE, as it is now the only way to remove nodes in Quine.
  • More standing query options are available on the Streams page, so building and tuning standing queries takes fewer trips to the API.
  • Result cards from quick queries on the Exploration UI are titled by quick query and node, so comparing several nodes side by side stays easy.

Security:

  • The Jackson JSON library is updated to fix two denial-of-service vulnerabilities.
  • Avro ingest is hardened against malformed input, so a bad record cannot exhaust memory.

Bugfixes:

  • Long-running API requests, such as large Cypher queries, now run to completion and return their results in a single call.
  • Graphs on Cassandra and Amazon Keyspaces recover on their own from incomplete snapshots, and paced snapshot writes protect storage during heavy load, keeping the graph available without operator intervention.
  • Historical queries in the Exploration UI honor the chosen time in every connection mode, and links can open the Exploration UI at a specific point in time, so a moment worth investigating can be shared with a URL.
  • Metrics stay accurate when ingests and standing queries are deleted and recreated under the same name, so dashboards and alerts reflect what is actually running.

Updates:

  • AWS SDK to 2.55.8
  • Amazon Kinesis client to 3.5.3
  • Apache Pekko to 1.7.0
  • API framework (Tapir) to 1.13.32
  • endpoints4s API libraries (OpenAPI 5.1.0, JSON Schema 1.13.0 and 2.7.0, Pekko HTTP server 2.1.0)
  • Circe JSON library to 0.14.17
  • Cats Effect to 3.7.1
  • Jackson JSON library to 2.22.3
  • Apache Avro to 1.12.2
  • Google Protocol Buffers to 4.36.2
  • Square Wire to 6.4.7
  • OpenTelemetry API to 1.66.0
  • Logback logging library to 1.6.5
  • SLF4J commons-logging bridge to 2.0.20
  • Dropwizard Metrics to 4.2.40
  • TLS configuration library (Ayza) to 10.1.0
  • jnr-posix native bindings to 3.2.4
  • Command-line parser (scopt) to 4.2.0
  • Scala parser combinators to 2.5.0
  • Scala Java Time (web UI date handling) to 2.7.0
Assets 3
Loading