hatchet-dev/hatchet v0.59.5

v0.59.5 - Priority queues, multiple concurrency keys, streaming improvements

on GitHub

What's Changed

New Features

• We're excited to announce SDK-side support for priority queues, which you can read more about here. With this feature, you can set the priority when triggering a task and also set a default priority for the task. For example, in the Python SDK:

  low_prio = priority_workflow.run_no_wait(
      options=TriggerWorkflowOptions(
          ## 👀 Adding priority and key to metadata to show them in the dashboard
          priority=1,
          additional_metadata={"priority": "low", "key": 1},
      )
  )
  
  high_prio = priority_workflow.run_no_wait(
      options=TriggerWorkflowOptions(
          ## 👀 Adding priority and key to metadata to show them in the dashboard
          priority=3,
          additional_metadata={"priority": "high", "key": 1},
      )
  )

• All SDKs now support setting multiple concurrency keys at both the workflow and the task level. Instead of passing a single concurrency strategy to a workflow, you can now pass an array of multiple strategies. Hatchet will move each task from one concurrency queue to the next before finally enqueueing the task for assignment on workers.

• The latest release adds performance improvements and stability for streaming features -- specifically, custom event streams now have stronger ordering guarantees, and awaiting a child workflow result is significantly faster.

Bug fixes

• Additional metadata is now respected on bulk replays and cancellation (#1503 and #1502)
• Python SDK version v1.6.1 fixes a critical issue related to reconnections with the Hatchet engine (#1557)
• Typescript SDK version 1.3.1 includes fixes for namespaces (#1549), fixes for typing on durable tasks (#1539)

Full Changelog: v0.58.1...v0.59.5