Reader function refactor #1979
Reader function refactor #1979
Conversation
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## master #1979 +/- ##
==========================================
+ Coverage 89.66% 89.68% +0.01%
==========================================
Files 876 877 +1
Lines 32337 32362 +25
==========================================
+ Hits 28996 29024 +28
+ Misses 3341 3338 -3
☔ View full report in Codecov by Sentry. |
| std::unique_ptr<ReaderMorsel> getParallelMorsel(); | ||
|
|
||
| inline void lock() { mtx.lock(); } | ||
| inline void unlock() { mtx.unlock(); } | ||
| inline common::row_idx_t& getNumRowsRef() { return std::ref(numRows); } | ||
|
|
||
| static void appendToArrayVectors(std::vector<arrow::ArrayVector>& arrayVectors, |
There was a problem hiding this comment.
The names of these two functions appendToArrayVectors and appendArrowArraysToVectors are quite confusing. I really can't differentiate them easily as they read so similarly...
For appendToArrayVectors, I think you mean appendArraysToVectors?
and for appendArrowArraysToVectors, maybe extractArraysToVectors?
There was a problem hiding this comment.
No, appendToArrayVectors means append data from valueVector to vector<arrow::ArrayVector>
appendArrowArraysToVectors means append data from ArrayVector to arrow vector
src/storage/copier/reader_state.cpp
Outdated
| } | ||
| } | ||
|
|
||
| arrow::ArrayVector ReaderSharedState::extractArrayVectorToAppend( |
There was a problem hiding this comment.
This is a part I'm a bit confused. Isn't the slicing and append logic the same for all arrays within the same set of batches? Why do we perform them repeatedly column by column? It seems if we don't organize data as per ArrayVector, instead per DataChunk (std::vector<std::shared_ptr<DataChunk>>), we can simplify this. Let's discuss a bit more on this, maybe I'm missing something.
There was a problem hiding this comment.
Changed the slicing algorithm
No description provided.