Skip to content

Release 2.14.12
Compare
Choose a tag to compare
@serinamarie serinamarie released this 20 Dec 22:59
· 2501 commits to main since this release
2.14.12
5a2482d
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

This release brings some exciting enhancements and fixes!

Enhanced UI Customization 📆

  • Now featuring relative time spans like "Past 7 days" for better insight into current and upcoming flow runs across the UI.
  • Flow Runs Page: Enhanced with customizable date and time filters. Save custom filters with specific date ranges for tailored views
  • Flows Page: Updated with similar date and time filtering capabilities

See a demonstration here!
short loom demo

Get type-checked input from humans in the loop

Human-in-the-loop flows just got an upgrade. You can now pause or suspend a flow and wait for type-checked input. To get started, declare the structure of the input data using a Pydantic model, and Prefect will render a form dynamically in the UI when a human resumes the flow. Form validation will ensure that the data conforms to your Pydantic model, and your flow will receive tbe input.

image

Prefect's new RunInput class powers this experience. RunInput is a subclass of Pydantic's BaseModel. Here's an example of a RunInput that uses dates, literals, and nested Pydantic models to show you what's possible:

class Person(RunInput):
    first_name: str
    last_name: str
    birthday: datetime.date
    likes_tofu: bool
    age: int = Field(gt=0, lt=150)
    shirt_size: Literal[ShirtSize.SMALL, ShirtSize.MEDIUM, ShirtSize.LARGE,
                        ShirtSize.XLARGE]
    shirt_color: Literal["red", "blue", "green"]
    preferred_delivery_time: datetime.datetime
    shipping_address: ShippingAddress
    billing_address: BillingAddress | SameAsShipping = Field(
        title="", default_factory=SameAsShipping
    )

Check out our guide on how to create human-in-the-loop flows to learn more!

Enhancements and Fixes 🛠️

  • Fixed environment variable parsing during deployment and prompt and role assignment in the ContainerInstanceProvisioner
  • Enhanced dashboard header to adapt to various tag and date input sizes
  • Addressed an error encountered when deploying flows with options

See the release notes for details!

Assets 2
Loading