lidavidm commented on code in PR #13364:
URL: https://github.com/apache/arrow/pull/13364#discussion_r895709083
##########
cpp/src/arrow/array/data.h:
##########
@@ -242,6 +245,127 @@ struct ARROW_EXPORT ArrayData {
std::shared_ptr<ArrayData> dictionary;
};
+/// \brief A non-owning Buffer reference
+struct ARROW_EXPORT BufferRef {
+ // It is the user of this class's responsibility to ensure that
+ // buffers that were const originally are not written to
+ // accidentally.
+ uint8_t* data = NULLPTR;
+ int64_t size = 0;
+ // Pointer back to buffer that owns this memory
+ const std::shared_ptr<Buffer>* owner = NULLPTR;
+};
+
+/// \brief EXPERIMENTAL: A non-owning ArrayData reference that is cheaply
+/// copyable and does not contain any shared_ptr objects. Do not use in public
+/// APIs aside from compute kernels for now
+struct ARROW_EXPORT ArraySpan {
+ const DataType* type;
+ int64_t length = 0;
+ mutable int64_t null_count = kUnknownNullCount;
+ int64_t offset = 0;
+ BufferRef buffers[3];
+
+ ArraySpan() = default;
+
+ explicit ArraySpan(const DataType* type, int64_t length) : type(type),
length(length) {}
+ explicit ArraySpan(const ArrayData& data) { SetMembers(data); }
+ explicit ArraySpan(const Scalar& data) { FillFromScalar(data); }
+
+ /// If dictionary-encoded, put dictionary in the first entry
+ // TODO(wesm): would a std::unique_ptr<vector<...>> be better?
Review Comment:
AFAIK SmallVector will switch to heap storage automatically, but now that I
think about it we can't use it here since then we'd have an infinitely
recursive type
--
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: github-unsubscr...@arrow.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
