jk-shah commented on issue #41346: URL: https://github.com/apache/beam/issues/41346#issuecomment-6934905846
Thanks for picking this up @Abhi2939! Yes — going with your first option (following `self._exception_level` and skipping `self._table.read_row(...)`) is definitely the right behavior. Nullable join keys (e.g. anonymous events with `user_id=None`) are common in left-join enrichment pipelines, so always raising would break pipelines configured with `ExceptionLevel.WARN` (the default) or `ExceptionLevel.QUIET`: Two small related details in `bigtable.py` to keep in mind while you're in there: 1. For `KeyError`: when `row_key_fn` is passed instead of `row_key`, `self._row_key` defaults to `""` — so scoping the `KeyError` catch specifically around `request_dict[self._row_key]` (rather than the `self._row_key_fn(request)` call) avoids turning a `KeyError` inside a custom `row_key_fn` into `"row_key not found..."`. 2. In `get_cache_key` (line 168), `"%s: %s" % (self._row_key, val)` maps both `None` and the literal string `"None"` to the same Redis cache key (`"user_id: None"`), in case you want to keep `None` distinct there too. Your test plan sounds great! -- 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]
