Skip rewriting internal pcollection coders in a stage #34655
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Prior to the PR, unknown coders were potentially replaced (e.g., length-prefixed) for every transform within a processing stage. This approach, however, caused issues in Java pipelines (like #32931), where worker nodes required the original coder (e.g.,
SchemaCoder
) to correctly relay data between internal transforms.An observation driving this change is that only the input and output coders of a stage are critical for the runner, as actual serialization/deserialization occurs at these stage boundaries. Coders for internal transforms should be irrelevant to the runner's processing and will now be kept intact. This minimizes potential side effects from modifying the pipeline graph and ensures internal data flow remains correct.
#fixes #32931