Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
This is something of a mega tracking issue. Once the dataset is reading asynchronously we can start converting the readers to be asynchronous. The primary goal is to keep blocking reads off the CPU thread pool. So "asynchronous" in this case means using the IOContext instead of the CPU thread pool.
Attachments
Issue Links
- contains
-
ARROW-1009 [C++] Create asynchronous version of StreamReader
- Open
-
PARQUET-1993 [C++] Expose when prefetching completes
- Resolved
1.
|
[C++] Add asynchronous read to ipc::RecordBatchFileReader | Resolved | David Li |
|
||||||||
2.
|
[C++] Add asynchronous read to parquet::arrow::FileReader | Resolved | David Li |
|
||||||||
3.
|
[C++] Add asynchronous read to streaming CSV reader | Resolved | Weston Pace |
|
||||||||
4.
|
[C++] Get rid of MakeIteratorGenerator | Resolved | David Li |
|