Zhen Li comments

Results 116 comments of


                                            Zhen Li

[VL] Respect prefix when register functions and convert function name

I encounter the same issue, so any guide on how to solve it? Thanks. https://github.com/apache/incubator-gluten/pull/5150#issuecomment-2024633762

[VL] Respect prefix when register functions and convert function name

I see, thanks your info.

[VL] Not fallback for function spark_partition_id and monotonically_increasing_id

Acked, let me check.

BaseVector::hashValueAt is not consistent with BaseVector::equalValueAt for NaNs

> Hash computation needs to be fixed in multiple places. > > * BaseVector::hashValueAt > * VectorHasher::hashValues > * RowContainer::hashTyped > * any other place that uses folly::hasher Spark performs...

BaseVector::hashValueAt is not consistent with BaseVector::equalValueAt for NaNs

Thanks. I check Spark `in` function also produces wrong result for NaN values.

BaseVector::hashValueAt is not consistent with BaseVector::equalValueAt for NaNs

Hello @mbasmanova and @Yuhta , May I ask what's the next step for this issue? It seems like the correct behavior for Spark and Presto is completely different. Spark expects...

BaseVector::hashValueAt is not consistent with BaseVector::equalValueAt for NaNs

Thanks. Get it, so the difference is only related to +/-0.0.

Add levenshtein Spark function

@rui-mo and @PHILO-HE , could you help check this PR? Thanks.

Improve 'insertForJoin' function performance by group prefetching

Verified with join fuzzer for 30 mins. ``` W0508 12:12:09.084839 3634699 JoinFuzzer.cpp:1232] ==============================> Done with iteration 9585 [==========] Running 0 tests from 0 test suites. [==========] 0 tests from 0...

Add rint Spark function

cc @rui-mo and @PHILO-HE , could you help check? Thanks.