[ https://issues.apache.org/jira/browse/LUCENE-2793?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13045354#comment-13045354 ]
Michael McCandless commented on LUCENE-2793: -------------------------------------------- Patch looks great! Comments: * I think IOCtx should have ctor taking only OneMerge, which would set the OneMerge and set the context as Merge? * Likewise, a ctor taking a SegmentInfo to mean context = Flush? * And finally a default ctor that maps to Other (or Default or Unknown or Unspecified or something) * You don't need to create the IOCtx in IW.maybeMerge? * Maybe we want a "readOnce" boolean in IOCtx? When we read del docs, norms, terms index, doc values, segments file / segments.gen, we would set this? (And UnixDir would send eg NO_REUSE down to the OS). * I think we'll need a NativeMMapDir as well as NativeDir (or NativeUnix/WindowsDir), because mmap can also take flags giving hints about access patterns. I'll open a new issue... * Why does SegmentCoreReaders hang onto the IOCtx? Seems like classes shouldn't hang onto it... (also: PreFlexFields). * Hmm.... does createOutput even need an IOCtx...? What would a dir do with this? I suppose if it's a merge and we had io prioritization (someday) we could set lower prio... OK let's keep it. * I think Codec.fieldsProducer/Consumer should take an IOCtx? * Still need to fix IW's ReaderPool to key off of IOCtx.Context plus the info. Maybe put a // nocommit in there so we remember...? Eg, where you commented out the readBufferSize = ... inside ReaderPool.get is a good place. > Directory createOutput and openInput should take an IOContext > ------------------------------------------------------------- > > Key: LUCENE-2793 > URL: https://issues.apache.org/jira/browse/LUCENE-2793 > Project: Lucene - Java > Issue Type: Improvement > Components: core/store > Reporter: Michael McCandless > Assignee: Varun Thacker > Labels: gsoc2011, lucene-gsoc-11, mentor > Attachments: LUCENE-2793.patch, LUCENE-2793.patch, LUCENE-2793.patch, > LUCENE-2793.patch, LUCENE-2793.patch, LUCENE-2793.patch > > > Today for merging we pass down a larger readBufferSize than for searching > because we get better performance. > I think we should generalize this to a class (IOContext), which would hold > the buffer size, but then could hold other flags like DIRECT (bypass OS's > buffer cache), SEQUENTIAL, etc. > Then, we can make the DirectIOLinuxDirectory fully usable because we would > only use DIRECT/SEQUENTIAL during merging. > This will require fixing how IW pools readers, so that a reader opened for > merging is not then used for searching, and vice/versa. Really, it's only > all the open file handles that need to be different -- we could in theory > share del docs, norms, etc, if that were somehow possible. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org