[ https://issues.apache.org/jira/browse/MAPREDUCE-326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12829718#action_12829718 ]
Doug Cutting commented on MAPREDUCE-326: ---------------------------------------- > So, if the input format runs in the user process, this doesn't really add > anything to language support until we have quality DFS clients in all the > major languages, right? Applications could use a 'dfs cat' tool to read their input, or mount HDFS using fuse or a native kernel driver. Or folks might not be using HDFS, but some other FS that already has good support in their language. This is a possible long-term direction, not an immediate plan. It's useful to consider long-term directions as we decide what to do near-term. I think a low-level binary API is consistent with language-independence, a long-term goal. I don't expect we'll have a single patch for this issue that will get us all the way there. This issue is only about adding a low-level binary API. Do folks see another way to achieve real language independence, where both mapreduce clients and servers may be implemented in other languages? Do folks even agree that such language-independence should be a long-term goal? Or should we rather continue to give Java a privileged status in the architecture long-term? > The lowest level map-reduce APIs should be byte oriented > -------------------------------------------------------- > > Key: MAPREDUCE-326 > URL: https://issues.apache.org/jira/browse/MAPREDUCE-326 > Project: Hadoop Map/Reduce > Issue Type: Improvement > Reporter: eric baldeschwieler > > As discussed here: > https://issues.apache.org/jira/browse/HADOOP-1986#action_12551237 > The templates, serializers and other complexities that allow map-reduce to > use arbitrary types complicate the design and lead to lots of object creates > and other overhead that a byte oriented design would not suffer. I believe > the lowest level implementation of hadoop map-reduce should have byte string > oriented APIs (for keys and values). This API would be more performant, > simpler and more easily cross language. > The existing API could be maintained as a thin layer on top of the leaner API. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.