pranavshuklaa commented on PR #1091:
URL: https://github.com/apache/flink-agents/pull/1091#issuecomment-6808539503
Thanks for the detailed review, @wenjin272 ! Hope this isn't taking too
much of your bandwidth. Thanks for the help 🙌
**Classpath failure cleanup (blocking):** `extractClasspathFromJars` uses
main's try/catch (IOException | RuntimeException | Error) again: it closes the
`Materialized` and rethrows. A failing later JAR no longer leaves the temp dir
and shutdown hook behind. The regression test
`extractClasspathFromJarsCleansUpWhenALaterJarFails` uses a valid first JAR and
a failing second JAR. I also gave `extractZipSafely(Path, Limits)` `main`'s
exact cleanup.
**`SkillOptions`:** the four options now live in `SkillOptions` in the
skills API module: Java `org.apache.flink.agents.api.skills.SkillOptions` and
Python `flink_agents.api.skills.SkillOptions`, with the same names, keys, types
and defaults. They are gone from `AgentConfigOptions` in both languages. The
extraction keys are now
`skill.source.max-extract-{entry-bytes,total-bytes,entries}`, and
`skill.source.url.max-download-bytes` is unchanged. The old names were never
released, so I didn't add aliases. The docs and the Java/Python config parity
check (`check_java_python_config_options_parity.py`) now cover `SkillOptions`.
**Tests:** I merged the duplicate download-cap tests in Java and Python,
dropped the redundant Java happy-path extraction test, and added small helpers
for the limits, the streaming server and the temp-dir cleanup checks. Every
distinct scenario is kept. The control-case repositories are now closed, and
the forged-size docstring is updated.
--
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]
