hy00nc commented on a change in pull request #222: [NEMO-350] Implement
Off-heap SerializedMemoryStore & [NEMO-384] Implement
DirectByteBufferInputStream for Off-heap SerializedMemoryStore
URL: https://github.com/apache/incubator-nemo/pull/222#discussion_r296088315
##########
File path:
common/src/main/java/org/apache/nemo/common/DirectByteBufferOutputStream.java
##########
@@ -140,38 +140,45 @@ public void write(final byte[] b, final int off, final
int len) {
final int arraySize = pageSize * (dataList.size() - 1) +
lastBuf.position();
final byte[] byteArray = new byte[arraySize];
int start = 0;
- int byteToWrite;
- for (final ByteBuffer temp : dataList) {
+ for (final ByteBuffer buffer : dataList) {
// ByteBuffer has to be shifted to read mode by calling
ByteBuffer.flip(),
// which sets limit to the current position and sets the position to 0.
// Note that capacity remains unchanged.
- temp.flip();
- byteToWrite = temp.remaining();
- temp.get(byteArray, start, byteToWrite);
+ final ByteBuffer dupBuffer = buffer.duplicate();
Review comment:
It does not literally copy the contents but **share** the same content of
the `ByteBuffer`, while keeping the limit and position independent from the
original one. `ByteBuffer` has a position and limit that it uses to write and
read the data. The important thing is that they are automatically incremented
and decremented by read/write(get/put). Every time we re-read the data or write
additional data after reading, etc., it is too complicated to set the position
and limit to the original one again. Sharing the same position and limit will
cause these errors, so it is best to keep independent values of position and
limit. This is the most tricky part of using `ByteBuffer`. Here we assumed that
this function is intended for reading the data from the start. I will comment
on that. 👍
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
