Wes McKinney
Wes McKinney
@pitrou there's a doctest failure here unrelated to this PR, seems to be caused by the 9.0.0 to 10.0.0 version bump (which made the metadata 1 byte bigger): ``` 2022-08-03T17:03:42.8027721Z...
I think the reason is the change from 9.0.0 to 10.0.0 which added one byte, here is the doctest ``` >>> pq.read_metadata('example.parquet') created_by: parquet-cpp-arrow version ... num_columns: 2 num_rows: 3...
I patched this in https://github.com/apache/arrow/pull/13790
I think this can be merged once the build is green if the changes look okay.
@pitrou this isn’t a micro-optimization imho — when 1% of your binary size is concerned with inline shared_ptr code in this narrow part of the library that speaks to a...
Any objections to merging this?
I do not think the implementation is less robust — the only change is that ephemeral `DataType` instances cannot be used to instantiate `InputType` without creating a TypeMatcher that wraps...
I've done a little exploration locally and concluded that `shared_ptr` versus `const DataType*` has no impact on the performance of `Function::DispatchExact` (at least on recent clang), which confirms the general...
I had some personal stuff come up in August that got in the way of completing this work -- I will pick this up hopefully sometime this month (September) and...
@ursabot benchmark please