----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/491/#review329 -----------------------------------------------------------
note that in this case - the maps used by different threads will almost surely not be accessed at the same time (because they are getting results back from the filesystem one at a time). I am a great fan of concurrent maps - but in this particular case - a synchronized map would likely be lower overhead. since we can't go wrong with either approach - i find it better to leave such things to the programmer. i also don't like using another thread unless it's warranted. surely there's some perf hit - why take it? Siying - can we avoid the repeated creation of a JobConf object? I realize this is older code - but this does have some overhead (since the creation does a clone of the underlying properties) - it would be good if we can get rid of it. - Joydeep On 2011-03-11 15:03:34, Carl Steinbach wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/491/ > ----------------------------------------------------------- > > (Updated 2011-03-11 15:03:34) > > > Review request for hive. > > > Summary > ------- > > Review request for HIVE-2051. > > > This addresses bug HIVE-2051. > https://issues.apache.org/jira/browse/HIVE-2051 > > > Diffs > ----- > > trunk/ql/src/java/org/apache/hadoop/hive/ql/exec/Utilities.java 1080771 > > Diff: https://reviews.apache.org/r/491/diff > > > Testing > ------- > > > Thanks, > > Carl > >
