bkietz commented on code in PR #14386:
URL: https://github.com/apache/arrow/pull/14386#discussion_r994747026
##########
cpp/src/arrow/compute/exec.cc:
##########
@@ -156,15 +156,23 @@ Result<ExecBatch> ExecBatch::Make(std::vector<Datum>
values) {
Result<std::shared_ptr<RecordBatch>> ExecBatch::ToRecordBatch(
std::shared_ptr<Schema> schema, MemoryPool* pool) const {
+ if (static_cast<size_t>(schema->num_fields()) > values.size()) {
+ return Status::Invalid("ExecBatch::ToTRecordBatch mismatching schema
size");
+ }
ArrayVector columns(schema->num_fields());
for (size_t i = 0; i < columns.size(); ++i) {
const Datum& value = values[i];
if (value.is_array()) {
columns[i] = value.make_array();
continue;
+ } else if (value.is_scalar()) {
Review Comment:
RecordBatch requires Array columns. ExecBatch also supports for Scalars,
which indicate a column with constant value. Such constant columns arise
frequently (for example when the dataset has been partitioned on a column so a
batch read from `cyl=4/dat.parquet` has a constant `cyl` column) and many
kernels can take directly accept a scalar argument (for example the arithmetic
kernels can add a scalar to an array), saving memory and processor overhead of
broadcasting to Array.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
