kaxil commented on code in PR #74694:
URL: https://github.com/apache/airflow/pull/74694#discussion_r4271017413
##########
providers/common/ai/docs/operators/llm_sql.rst:
##########
@@ -59,6 +59,13 @@ the actual column names and types:
With Object Storage
-------------------
+.. note::
+
+ Experimental: ``datasource_config`` can change or be removed in a minor
release of this
Review Comment:
The stable `@task.llm_sql` row in `stability.rst` (line 100) still names
`datasource_config` as a way to give the model the schema, and the
`@task.llm_schema_compare` row (line 112) promises introspecting
`data_sources`. Those rows need an "experimental; see below" like
`decision_policy` has, plus an entry in the experimental table, or this note
and the stability page contradict each other. Same for the new note in
`llm_schema_compare.rst`.
##########
providers/common/ai/docs/changelog.rst:
##########
@@ -25,6 +25,14 @@
Changelog
---------
+.. note::
+ ``airflow.providers.common.ai.toolsets.logging.LoggingToolset`` (shipped in
0.10.0) is
Review Comment:
`LoggingToolset` has been public since 0.1.0 (it is in `toolsets/logging.py`
in every `providers-common-ai` tag from 0.1.0 on), so "shipped in 0.10.0" reads
as if it was new there. Dropping the parenthetical, or saying "public since
0.1.0", avoids that.
##########
providers/common/ai/docs/frameworks/index.rst:
##########
@@ -130,9 +130,13 @@ Tested versions
- Pins OpenTelemetry at 1.42.1 or lower, below the version in Airflow's
constraints
file. Airflow itself accepts 1.42.1.
-These are the versions the adapter tests have been run against. CI does not
run those
-tests: its environment has ``mcp`` 2.2 and OpenTelemetry 1.44, which Strands
and ADK
-exclude. They run in an environment with the framework installed.
+These are the versions the adapter tests have been run against. The main test
matrix
+skips those tests: its environment has ``mcp`` 2.2 and OpenTelemetry 1.44,
which Strands
+and ADK exclude. A separate CI job installs one framework at a time and runs
them, on
Review Comment:
Outside canary runs, this job holds every package in the image with
`--override` (`run_agent_framework_tests.sh`), so Strands' and ADK's own caps
are overridden, and it installs the newest framework release rather than the
versions in the table above. One sentence saying so would stop this reading as
CI covering those versions in the environment a user gets.
##########
providers/common/ai/docs/toolsets/logging.rst:
##########
@@ -15,19 +15,12 @@
specific language governing permissions and limitations
under the License.
-Tool call logging: ``LoggingToolset``
-=====================================
+Tool call logging
+=================
-.. note::
-
- Experimental: this can change or be removed in a minor release of this
provider.
- See :ref:`howto/stability`.
-
-:class:`~airflow.providers.common.ai.toolsets.logging.LoggingToolset` is a
-``WrapperToolset`` that intercepts ``call_tool()`` to log each tool invocation
-in real time. ``AgentOperator`` applies it automatically (see
-``enable_tool_logging``) through
-:class:`~airflow.providers.common.ai.toolsets.logging.ToolLoggingCapability`.
+:class:`~airflow.providers.common.ai.toolsets.logging.ToolLoggingCapability`
wraps the
+assembled function toolset so that each tool invocation is logged in real time.
+``AgentOperator`` applies it automatically (see ``enable_tool_logging``).
Review Comment:
This now repeats the paragraph that starts "`AgentOperator` adds
`ToolLoggingCapability` automatically" below. The page also lost its only
example for an agent you build yourself; the `Agent(...,
capabilities=[ToolLoggingCapability()])` from the changelog would fit where the
`LoggingToolset` block was.
##########
providers/common/ai/docs/stability.rst:
##########
@@ -246,10 +246,6 @@ Everything this provider ships that is not in the table
above is experimental.
``capture_content``; :doc:`observability`)
- Span names and attributes come from Pydantic AI's instrumentation and
the
OpenTelemetry GenAI conventions, which are still in development.
- * - :class:`~airflow.providers.common.ai.toolsets.logging.LoggingToolset`
used directly
Review Comment:
Since the changelog now sends people to `ToolLoggingCapability` for agents
they build themselves, could this row point at it instead of going away? With
this row and the note on `toolsets/logging.rst` both removed, the capability is
experimental only through the catch-all sentence, and the logging page no
longer says so.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
