Well, Bhavani found a bug! I'll be putting in a Jira. It's quite an edge case:
1) occurs in delta deserialization 2) happens only when there's a set of modifications to existing feature structures done by the remote service 3) happens only when the sequence of modifications recorded by the delta cas mechanism records a modification to a feature structure that's before the last one modified 4) happens only when the feature structure being modified is exactly the previous one from the last one modified. 5) happens only when the feature structure being modified has a number of features that is shorter than the item being modified's offset The code was searching using Arrays.binarySearch over a range, and the ending element of the range is "exclusive", but the code thought it was "inclusive" - an off-by-1 error... Jira coming. -Marshall On 7/28/2015 1:21 PM, Marshall Schor wrote: > Hi, > > I've posted the UIMAJ SDK 2.8.1 rc1 release candidate. > > This version has mainly bug fixes for issues found as 2.8.0 got a bit wider > use. > > The list of changes in Jira: > > https://issues.apache.org/jira/issues/?jql=fixVersion%20%3D%202.8.1SDK%20AND%20project%20%3D%20UIMA > > The source and binary zip/tars and the Eclipse update site are staged to > http://people.apache.org/~schor/uima-release-candidates/uimaj-2.8.1-rc1/ > <http://people.apache.org/%7Eschor/uima-release-candidates/uimaj-2.8.1-rc1/> > > The Maven artifacts are here: > https://repository.apache.org/content/repositories/orgapacheuima-1067 > > The SVN tags are here: > http://svn.apache.org/repos/asf/uima/uimaj/tags/uimaj-2.8.1/ > > and for the Eclipse Update Site: > http://people.apache.org/~schor/uima-release-candidates/uimaj-2.8.1-rc1/eclipse-update-site/uimaj/ > <http://people.apache.org/%7Eschor/uima-release-candidates/uimaj-2.8.1-rc1/eclipse-update-site/uimaj/> > > See http://uima.apache.org/testing-builds.html for suggestions on how to test > release candidates. > > Please vote on release: > > [ ] +1 OK to release > [ ] 0 Don't care > [ ] -1 Not OK to release, because ... > > Thanks. > > -Marshall > > > > >
