On 3/30/2016 8:49 AM, Nicholas Knize wrote: > It appears all 6.0 blocker issues are coming to a close. If there are > no objections I will begin the RC process today.
I debated with myself about whether or not to even send this message. Jack Krupansky mentioned it, so I'm going ahead. I think that LUCENE-6917 should be reverted in branch_6x and branch_6_0, but left in place in master. Alternately, we could delay the removal of the legacy types until 8.0. I'm not trying to suggest that the dimensional types aren't good, or that the older types shouldn't be removed eventually. I'm just saying that we need to push it back one major release -- allow both sets of classes to exist as peers in 6.x, with the older classes deprecated in 7.0and removed in 8.0. Side issue: If we do revert in 6x, the class rename has potential to cause pain for backports from master to stable. I'm not sure a rename is necessary -- isn't deprecation enough? This is of particular importance for Solr. As things stand right now, Solr 6.0 will ship with all its numeric types using deprecated Lucene code. This means that a new user who starts with an example config from Solr 6.0 will not be able to upgrade to Solr 7.0 without big config changes and a complete reindex, because the code required to read their index data will be gone. Thanks, Shawn --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
