Compare dictionary with primitive array in eq_dyn and neq_dyn
#2533
Conversation
| @@ -2173,45 +2270,28 @@ macro_rules! typed_dict_compares { | |||
| }}; | |||
| } | |||
|
|
|||
| /// Helper function to perform boolean lambda function on values from two dictionary arrays, this | |||
| /// version does not attempt to use SIMD explicitly (though the compiler may auto vectorize) | |||
| fn compare_dict_op<'a, K, V, F>( | |||
There was a problem hiding this comment.
As TypedDictionaryArray implements ArrayAccessor too, we can use compare_op to replace this compare_dict_op. So removing it here.
|
Benchmark runs are scheduled for baseline = 4949a3d and contender = 1d5656e. 1d5656e is a master commit associated with this PR. Results will be available as each benchmark for each run completes. |
| let left_iter = left.into_iter(); | ||
| let right_iter = right.into_iter(); | ||
|
|
||
| let result = left_iter |
There was a problem hiding this comment.
As pointed out by @pvsri #2564 (comment) this change introduces unsoundness, as dictionary indexes aren't always valid. I will have a think about how best to handle this
There was a problem hiding this comment.
Should we just call self.values.value instead of value_unchecked in ArrayAccessor.value_unchecked of TypedDictionaryArray? It might be less performance, though.
There was a problem hiding this comment.
No, because that will panic... I'm working on a fix - standby
Which issue does this PR close?
Closes #2535.
Rationale for this change
What changes are included in this PR?
Are there any user-facing changes?