sunxi92 opened a new pull request #2741: URL: https://github.com/apache/rocketmq/pull/2741
While reading the RocketMQ source code, I found that there are some areas that could be optimized 1.The getData(final long offset) function in the DledgerCommitLog.java is used to get data according to the offset. In Dledger, dividedCommitlogOffset is used to separate the old commitlog from dledger commitlog, so the getData(final long offset) function should judge if offset is in old commitlog or dledger commitlog. But getData(final long offset) function repeats the judgment of whether the offset is in old commitlog or dledger commitlog. So the if statement can be deleted. 2.The checkMessageAndReturnSize(ByteBuffer byteBuffer, final boolean checkCRC,final boolean readBody) function in the DledgerCommitLog.java is called when build the DispatchRequest. The same with getData function, it should judge whether the byteBuffer is in old commitlog or dledger commitlog.The following figure shows the data format in both modes，we can see that the first field is magic in dledger, the value of magic has two value: DLedgerMmapFileStore.CURRENT_MAGIC(1) or MmapFileList.BLANK_MAGIC_CODE(-1). In old commitlog, this field is TOTALSIZE and the value in this field must be gt 1(every commitlog will be free at least 8 bytes). So we can judge the bytebuffer from the first four bytes，if the value of the first four bytes is gt 1，then the bytebuffer is in old commitlog.  In addition，isInrecoveringOldCommitlog is used to indicate that it is the first time to load mixed commitlog, need to recover the commitlog. It has nothing to do with whether the byteBuffer is in old commitlog or dledger commitlog. So I think the if statement can be deleted. #2740 -- 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]
