Presumably this has been cut already. If this RC is sunk for some reason, please consider a resolution to HBASE-9549 for the next. This cleans up an inconsistency in the way we handle an empty/null column qualifier in our edge interfaces (thrift, rest, mapreduce). It all comes down to proper parsing in KeyValue#parseColumn. It is a breaking change in a rather subtle way for an edge case.
Thanks, Nick On Mon, Sep 16, 2013 at 11:17 PM, Stack <[email protected]> wrote: > Cutting second RC now. > St.Ack > > > On Mon, Sep 16, 2013 at 5:22 PM, Stack <[email protected]> wrote: > > > On Mon, Sep 16, 2013 at 5:13 PM, Devaraj Das <[email protected]> > wrote: > > > >> On Mon, Sep 16, 2013 at 5:10 PM, Stack <[email protected]> wrote: > >> > On Mon, Sep 16, 2013 at 1:34 PM, Devaraj Das <[email protected]> > >> wrote: > >> > > >> >> It seems like "move" in general has issues - HBASE-9525 and > HBASE-9480 > >> >> are other symptoms of the problem. Maybe, there are more... Maybe we > >> >> should spend a little time to review code and fix such issues? > >> >> > >> >> > >> > HBASE-9525 has been resolved already as has HBASE-9480. Jimmy is on > >> > HBASE-9514. He has a patch up for review. > >> > > >> > You thinking we NOT cut an RC before spending " a little time to > reivew > >> > code and fix such issues"? > >> > > >> > >> Now that Jimmy has fixed the issues (thanks, Jimmy!), I am personally > >> okay to let an RC happen now. > >> > > > > OK. > > > > There are still some issues to go in. Trying to verify they don't break > > stuff w/ local machine and navigating hadoopqa. > > > > HBASE-9538 "Handle post namespace snapshot files when checking for HFile > > V1" -- I'll do this one. > > HBASE-9510 "Namespace operations should throw clean exceptions" -- Enis > to > > commit > > > > Sergey has a commit for 0.96 that he is on. > > > > HBASE-9514 might make it too. > > > > St.Ack > > > > > > > > > > >
