> So, I guess the ReadRecordBatch function is intended to only work if the
> records were written by RecordBatchFileWriter, right?
The *StreamWriter and *FileWriter classes use identical code paths for
writing the IPC messages, the only difference is the preamble (the
magic number and padding) an
Hi Dimitri,
Thanks for that. I was going something like that. My hope was to use
StreamWriter to write the batch and then use ReadRecordBatch to read it
since it is more succinct and I know I only have one batch to read.
Here is the actual code I use
https://github.com/Paradigm4/accelerated_io_to
Hi Rares,
you use a different reader for the RecordBatch streams. See
arrow/ipc/ipc-read-write-test.cc:569-596 for the gist. Also, the second
argument to arrow::RecordBatch::Make takes the number of rows in the batch,
so you have to set it to 1 in your example.
See https://gist.github.com/alendi
Hi,
I'm writing a batch of records to a stream and I want to read them later. I
notice that if I use the RecordBatchStreamWriter class to write them and
then ReadRecordBatch function to read them, I get a Segmentation Fault.
On the other hand, if I use the RecordBatchFileWriter class to write the
