BryanCutler commented on a change in pull request #22305:
[SPARK-24561][SQL][Python] User-defined window aggregation functions with
Pandas UDF (bounded window)
URL: https://github.com/apache/spark/pull/22305#discussion_r241609712
##########
File path: python/pyspark/worker.py
##########
@@ -238,7 +284,8 @@ def read_udfs(pickleSer, infile, eval_type):
# In the special case of a single UDF this will return a single result
rather
# than a tuple of results; this is the format that the JVM side
expects.
for i in range(num_udfs):
- arg_offsets, udf = read_single_udf(pickleSer, infile, eval_type,
runner_conf)
+ arg_offsets, udf = read_single_udf(
+ pickleSer, infile, eval_type, runner_conf, udf_index=i)
Review comment:
Yeah, basically this
```
window_eval_type_str, remaining_type_str =
runner_conf['pandas_window_bound_types'].split(',', 1)
runner_conf['pandas_window_bound_types'] = remaining_type_str
window_eval_type = window_eval_type_str.strip().lower()
```
I'm not crazy about changing the conf inplace, but it wouldn't rely on any
particular udf indexing then. Maybe it would make more sense to check the eval
type before calling `read_single_udf`, process the conf and then send the
window_eval_type as an optional param to `read_single_udf`?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
