Update Solr to a more recent version

[ruebot]

We're running 4.10. Any reason we shouldn't move up? Should we go for Solr 6+?

@Islandora-CLAW/committers thoughts?

@Islandora-CLAW/sprinters let us know what you think too!

[MarcusBarnes]

I'm listening in on the committers call. PHP version support came up (http://php.net/eol.php). It would be a good idea to make sure we're building using versions that are not end-of-life or near end-of-life. Is such information available for the versions of Solr?

[acoburn]

I still use Solr 4.x for certain integration testing suites b/c they still provide a WAR file under that version, which makes certain things easier. AFAIK, the main query/update API has not changed in a very long time, so there's no problem with testing w/ 4.x, unless you're planning to use some of the newer, more advanced features (admin API, stand-alone deployment, etc, etc)

[ruebot]

@Islandora-CLAW/committers are y'all fine on staying on 4.x, or do we have a use case for a new feature? If we decide to stay where we're at, I'll make this issue resolved.

[DiegoPino]

@ruebot we really should explore 5.x or even better 6.x. Drupal Search API support for multilingual search works better with 5.x, 6.x allows per field update (remember whe mentioned this to cm harlow a few months ago?).

[ruebot]

@DiegoPino can you outline use cases for moving away from 4.x, and provide a version number we should move to? Once we have agreement on that, I'll go update the Vagrant version, and we'll develop against that.

[DiegoPino]

@ruebot crazy day, taking a bus to catch a plane now. May i do this tomorrow?

[ruebot]

@DiegoPino no rush on this. Take your time 😄

[qadan]

weirdly poking my head in here

I've been doing some work recently to try to get some of the features of Solr 5.0+ hack-ported into 4.10.x and it's been a real pain in the rear end. Specifically:

• The DateRangeField, which was added in 5.0. I'll be putting a pull in to Islandora Transforms shortly to make date range faceting work in in 4.10.x but it's really awful. Being able to facet on objects that were "created between xxxx-yyyy" or "published after zzzz" would be a huge improvement to searching.
• Query-time multi-word synonym expansion, without which Solr synonyms are basically useless. This actually doesn't even exist yet, but I've been following the related JIRA issue and have the patch working - just, the only way they'll accept it upstream is as part of a branch that's in active development, which now precludes 4.x. I'm going to put some work into porting the patch forward, but it seems like as far as they're concerned, it'll never be accepted into 4.x.

Additionally, something I miss in 4.10.x is the massive improvements to the Schema API in 5.0+. The place it's at now would make some major Solr configuration doable from the Drupal side without having to touch the configs - it could even be baked into field configuration, for example.

I'd say if you're looking for a target version, 5.0 provides a couple fantastic features, and the next major dev versions will too if I can get that patch assembled.

[ruebot]

@qadan thanks! That sounds like a really solid use case for moving to at least 5.x.

[ruebot]

Completely forgot that Solr 5.0 and up cannot be deployed in Tomcat: https://wiki.apache.org/solr/WhyNoWar

Hacking notes...

• https://cwiki.apache.org/confluence/display/solr/Solr+Start+Script+Reference
• https://cwiki.apache.org/confluence/display/solr/Taking+Solr+to+Production
• Need to open a port; 8983
• Need a deploy path; /opt? /usr/local/solr?
• Need a solrconfig.xml
• Need a schema.xml

[ruebot]

PR: #407

[dannylamb]

Resolved with 01c4dd0