Collective.solr

------------------

- Extend the wildcard search term manipulation to do Unicode to ascii folding,
to keep up with the default field settings of the text field.
[hannosch, mj]

------------------

- Don't treat search terms ending in numbers as `simple`, as Solr doesn't deal
with wildcard searches for numbers.
[hannosch]

- Include CMFCore's `permissions.zcml`.
[witekdev, hannosch]

------------------

* Fixed bug in `extender.searchwords` indexer - terms need to be lowercased
explicitly.
[hannosch]

------------------

* Fixed handling of intra-word hyphens to be taken literally instead of being
interpreted as syntax for text fields.
[hannosch]

* Explicitly require Plone 4.1 / Zope 2.13.
[hannosch]

* Depend on the new c.indexing 2.0a2.
[hannosch]

* Added an `archetypes.schemaextender` dependency and register two fields for
all objects providing `IATContentType`. `showinsearch` is a boolean field that
can be used to hide specific content items from search results. `searchwords`
is a lines field, which lets you specify words that an object should be found
under.
[hannosch]

* Standardize on `solr` as the i18n domain.
[hannosch]

------------------

* Adjust munin configs for query cache handlers to `c.r.solrinstance 3.5`
changes using `FastLRUCache`.
[hannosch]

* Added munin configs for the `/update/extract`, the direct update handler,
query cache size and warmup time, admin file requests used to get the
Solr schema and the searcher warmup time.
[hannosch]

* Added tests for splitting words on `:` and `-`.
[hannosch]

* Update example configuration to Solr 3.3.
[hannosch]

* Add `getRID` and `_unrestrictedGetObject` to our flare implementation.
[hannosch]

* Added documentation on setting up a master-slave configuration using the
`SolrReplication` support.
[hannosch]

* Adjust tests to work with latest `collective.recipe.solrinstance = 3.3` and
its new ICU-based text field.
[hannosch]

------------------

Upgrade notes
*************

* Changed the names of the indexes used to emulate the `path` index. You need
to adjust your schema and rename `physicalPath` to `path_string`,
`physicalDepth` to `path_depth` and `parentPaths` to `path_parents`. This
also requires a full Solr reindex to pick up the new data.
[hannosch]

Changes
*******

* Added `object_provides` index to example schema, as it's used in the
collection portlet to find collections.
[hannosch]

* Rewrote the `maintenance/sync` method for more performance, dropped the
optional `path` restriction from it and removed the `cache` argument. It
should be able to sync datasets in the 100,000 object range in the matter of
a couple minutes.
[hannosch]

* Changed the `maintenance/reindex` method to only flush data to Solr but not
commit after each batch. Instead we only commit once at the end. You should
configure auto commit policies on the Solr server side or `commitWithin`.
[hannosch]

* Adjusted the `mangleQuery` function to calculate extended path indexes from
the Solr schema instead of hardcoding `path`. If you have any additional
extended path indexes, you need to provide indexers with the same three
suffixes as we do ourselves in the `attributes` module for the `path` index
and add those to the Solr schema.
[hannosch]

* Added documentation on Java process, monitoring production settings and
include a number of useful munin plugin configurations.
[hannosch]

* Updated example config to include production settings and JMX.
[hannosch]

* Updated example config to collective.recipe.solrinstance 3.1 and Solr 3.2.
[hannosch]