[Data] Migrate Daft dependency from getdaft to daft

[jiangxt2]

Description

Migrate the Daft dependency in Ray's CI from the deprecated getdaft package
to the actively maintained daft package, and remove the now-obsolete PyArrow

= 14 version check in from_daft().

The getdaft package was deprecated (last release 0.5.0, Jan 2026). The Daft
project moved to the daft package on PyPI (latest: 0.7.15). The PyArrow >= 14
check was needed because old Daft referenced pa.PyExtensionType directly,
which was removed in PyArrow 14. Daft >= 0.7.0 fixed this with a hasattr
guard.

Related issues

Fixes #64178

Additional information

• test_daft_round_trip uses check_dtype=False because Ray returns
  Arrow-backed dtypes while Daft may return numpy dtypes after round-trip.
• CI lock files (requirements_compiled*.txt) will be updated on next compile.
• daft>=0.7.0 uses lower-bound constraint consistent with other ML data
  dependencies in Ray.

[gemini-code-assist]

Code Review

This pull request updates the integration with Daft by migrating from the deprecated getdaft package to daft>=0.7.0, updating documentation URLs to daft.ai, and removing the restriction that prevented from_daft from working with PyArrow 14 and later. Additionally, tests are updated to handle dtype differences and tensor column comparisons. Feedback suggests adding a defensive version check in from_daft to ensure daft>=0.7.0 is installed, preventing potential issues for users with older versions of the package.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

Since the deprecated getdaft package (version <= 0.5.0) also installs the daft module, users who still have the old package installed might encounter cryptic errors when using from_daft with PyArrow >= 14. To prevent this and ensure a smooth migration, we should add a defensive version check to verify that the installed daft package is indeed >= 0.7.0.

Suggested change

	"""
	"""
	import daft
	from packaging.version import parse as parse_version
	if parse_version(daft.__version__) < parse_version("0.7.0"):
	raise ImportError(
	f"ray.data.from_daft requires daft >= 0.7.0, but found {daft.__version__}. "
	"Please upgrade daft via 'pip install -U daft'."
	)

[jiangxt2]

Thanks for the suggestion! I've added a defensive version check in from_daft() to ensure daft >= 0.7.0 is installed. This prevents cryptic AttributeError for users with residual getdaft<=0.5.0 who haven't upgraded to the new package.

The check raises a clear ImportError with upgrade instructions if an older version is detected.

[dstrodtman]

LGTM for docs, assuming Ray Data team signs off on dependency updates and CI changes.