I suspect that what will happen is that the initial 5.0 of Solr will not
have a lot of draw based on features per se. Solr 4.10.x will become like
Solr 3.6.x - the stable, safe choice for another year or so. Then,
incrementally, the 5.x releases will gradually accumulate new features and
stability, while 4.10.x sits relatively stagnant, so that in a year or so
some 5.x will reach a critical mass of enhancements and stability and become
the default for new apps.
IOW, go ahead and release Solr 5.0 as soon as some committer has the time to
be RM. No real harm even if it has any serious issues since anybody with an
absolute requirement for stability will stay with 4.10.x anyway. Put another
way, release 5.0 whenever we would have released 4.11.
Then... we can sit back and watch to see what Elasticsearch and Heliosearch
do - will they leap immediately to 5.0, or stay with 4.10 for while, or
support both?
In my own bailiwick, DataStax Enterprise, we're waiting for Solr 4.10.x to
bake for another couple of months before we integrate it in our product, and
will probably want to see Solr 5.x bake for a good six to nine months before
we switch over to it. Those are just my own personal, tentative thoughts, no
commitment from myself or my client intended.
In any case, I enthusiastically look forward to seeing Solr 5.0 come out the
door before we know it!
-- Jack Krupansky
-----Original Message-----
From: Erick Erickson
Sent: Monday, October 13, 2014 5:39 PM
To: dev@lucene.apache.org
Subject: Next Solr release (5.0)
Hmmm, I'm thinking we need to start laying the groundwork on the
user's list for this maybe? It's a somewhat unusual "major" release in
that there's nothing scary-new compared to 4.10.
In the past, e.g. the 3.x->4.0 release was something I'd have been
reluctant to put into production without a good long lead time to
test; there was a _lot_ of new ground there! And there are
"interesting" company policies about whether they allow major
releases. Completely arbitrary restrictions, but they exist.
Do note that I'm _not_ advocating that we change the release process
or anything like that. What I _am_ wondering is if we should start a
discussion on the user's list to give people some time to adjust. In
which case having a good summary would be useful.
Or maybe it's just as well to leave it and expect a bunch of "what
does it all mean" questions when we announce the next release.
FWIW,
Erick
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
