Skip to content

views.md

Latest commit

 

History

History
64 lines (45 loc) · 1.69 KB

views.md

File metadata and controls

64 lines (45 loc) · 1.69 KB
Raw

Views

IceFrame exposes a ViewManager abstraction over Iceberg views.

Catalog support is not universal. Iceberg views are a catalog-level feature. REST catalogs that implement the view spec (Polaris, Tabular, Dremio) support them; PyIceberg's sql (SQLite) and memory catalogs do not. Calls against a catalog without view support raise an error from the catalog itself — IceFrame does not emulate views locally.

Creating a view

from iceframe import IceFrame, load_catalog_config_from_env

ice = IceFrame(load_catalog_config_from_env())

ice.create_view(
    "analytics.active_users",
    "SELECT id, name FROM analytics.users WHERE active = true",
)

replace=True replaces an existing view instead of failing:

ice.create_view("analytics.active_users", sql, replace=True)

Dropping a view

ice.drop_view("analytics.active_users")

Checking support before you rely on it

Because support varies, guard view usage rather than assuming it:

from iceframe import IceFrameError

try:
    ice.create_view("analytics.v", "SELECT 1")
except IceFrameError as e:
    print(f"This catalog does not support views: {e}")

Alternatives when your catalog has no views

  • DataFusion SQL (ice.query_datafusion(...)) runs SQL locally over one or more Iceberg tables without needing catalog view support.

  • The query builder composes reusable query fragments in Python:

    def active_users(ice):
    return ice.query("analytics.users").filter(col("active") == True)  # noqa: E712

See also