Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix side-effect of default options in Beam Operators #37916

Merged
merged 1 commit into from
Mar 5, 2024

Conversation

potiuk
Copy link
Member

@potiuk potiuk commented Mar 5, 2024

Some of the operators in Apache Beam had side effect that they modified detault options passed to it in constructor, and in case of xdist tests it had side effect that they could impact other test results (this happened recently in main). The default options are already set by the Dataflow mixin in execute method in all Beam operators, but in Python and Go operator the defaults are also set in the constructor.

Setting the defaults in mixin uses deepcopy to avoid such side effects.

This might be intended, so this PR rather than removing default settings in the constructor, also adds deepcopy in them and fixes resulting tests - removing the defaults in tests that do not have the defaults set in the constructor.


^ Add meaningful description above
Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragment file, named {pr_number}.significant.rst or {issue_number}.significant.rst, in newsfragments.

Some of the operators in Apache Beam had side effect that they
modified detault options passed to it in constructor, and in case
of xdist tests it had side effect that they could impact other
test results (this happened recently in main). The default options
are already set by the Dataflow mixin in execute method in all
Beam operators, but in Python and Go operator the defaults
are also set in the constructor.

Setting the defaults in mixin uses deepcopy to avoid such side effects.

This might  be intended, so this PR rather than removing default
settings in the constructor, also adds deepcopy in them and fixes
resulting tests - removing the defaults in tests that do not
have the defaults set in the constructor.
@potiuk potiuk merged commit c29d728 into apache:main Mar 5, 2024
58 checks passed
@potiuk potiuk deleted the fix-side-effect-in-beam-tests branch March 5, 2024 21:55
@potiuk
Copy link
Member Author

potiuk commented Mar 5, 2024

So - ideally - we could likely remove if we are sure it's fine, but even if we could, it should be a separate PR IMHO.

        self.pipeline_options.setdefault("labels", {}).update(
            {"airflow-version": "v" + version.replace(".", "-").replace("+", "-")}
        )

@Taragolis
Copy link
Contributor

Yeah, all logic around templated_fields should be moved into the execute method in any way, otherwise it will only work in some simple cases.

utkarsharma2 pushed a commit to astronomer/airflow that referenced this pull request Apr 22, 2024
Some of the operators in Apache Beam had side effect that they
modified detault options passed to it in constructor, and in case
of xdist tests it had side effect that they could impact other
test results (this happened recently in main). The default options
are already set by the Dataflow mixin in execute method in all
Beam operators, but in Python and Go operator the defaults
are also set in the constructor.

Setting the defaults in mixin uses deepcopy to avoid such side effects.

This might  be intended, so this PR rather than removing default
settings in the constructor, also adds deepcopy in them and fixes
resulting tests - removing the defaults in tests that do not
have the defaults set in the constructor.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants