Thanks Ted and Stack. We're fine on our end, but I'll go ahead and open a Jira and submit a patch for this anyway.
> On Feb 9, 2015, at 1:35 PM, Stack <[email protected]> wrote: > > On Mon, Feb 9, 2015 at 9:41 AM, Ian Friedman <[email protected]> wrote: > >> Hello! I'm in the process of upgrading our HBase 94 codebase to HBase 98. >> The org.apache.hadoop.hbase.filter.Filter class now has a field >> protected boolean reversed; which seems to, from my reading of the code >> that touches it, be a transient field set only when the Scan using the >> filter is itself reversed. It doesn't seem like it's supposed to be >> serialized/deserialized with the rest of the filter. However it is not >> marked transient, which trips up our serialization unittests... Is my >> reading wrong and this field should get serialized (despite being protected >> in a abstract base class) or should this field actually be transient? >> > > You are reading it correctly Ian. It should be marked transient. It is not > serialized when a Filter is serialized and apparently is not meant to be. > It is actually confusing what this flag is about. > > Do you need us to make a release with this set transient? (If upgrading, > what else can we do to make your life easier?) > > St.Ack
