[ https://issues.apache.org/jira/browse/HADOOP-11541?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14308438#comment-14308438 ]
Kai Zheng commented on HADOOP-11541: ------------------------------------ Hi [~zhz], Thanks for your comments. bq.What's the downside of converting to a unified format first? I'm thinking it's doing the real calculating and performance concerned, so I don't have much trick here. When we have more experience for implementing the following more raw coders, I guess we can have some idea about how to unifying them. bq.TestCoderBase needs better documentation Agree, will do. bq.What does {{getErasedIndexesForDecoding}} do? What is {{erasedDataIndexes}} for? To test recovering, we intentively erase some data units specified by {{erasedDataIndexes}}. {{getErasedIndexesForDecoding}} is to adjust the indexes to pass to raw coder to invoke the decode() function. Will have more comments for them. bq.{{testCoding}} can be in {{TestCoderBase}} {{TestCoderBase}} is not only for tests of raw coders, but also will be used for tests of block coders, which can be seen in the patch for HDFS-7662. Therefore I don't put raw coder specific codes in the base class. > Raw XOR coder > ------------- > > Key: HADOOP-11541 > URL: https://issues.apache.org/jira/browse/HADOOP-11541 > Project: Hadoop Common > Issue Type: Sub-task > Reporter: Kai Zheng > Assignee: Kai Zheng > Fix For: HDFS-EC > > Attachments: HADOOP-11541-v1.patch > > > This will implement XOR codes by porting the codes from HDFS-RAID. The coder > in the algorithm is needed by some high level codecs like LRC. -- This message was sent by Atlassian JIRA (v6.3.4#6332)