I would definitely steer clear from one record with multiple languages in the one record . . . otherwise it might be hard to avoid false positive matches on fields containing text from the wrong language.

Localization is much more interesting than translation, so make the assumption that for each record there may not be an equivalent translation. For example, while a site might be translated into two languages, each language may have a different set of pages, so not every page has a direct translated analogue.

Each record could only have text in one language and have fields indicating the language and relevant locale(s) so results can be filtered by language or locale.

result.facet(Solr1, "astring").values

I also want to create a collection of taglibs for constructing style-able search UI. For example:
<solr:facet field="genre"/>
<solr:facet field="year"/>
<solr:search autocomplete="true"/>
<solr:results pageSize="10"/>

You can override this also if you have an attribute in your domain named title.

And then further, you can define overrides for how it indexes fields by defining methods like indexSolr<attribute_name> like this:

def indexSolrTitle(doc) {
doc.addField(this.solrFieldName("title"), title?.toUpperCase())
}

This would also allow you to index the attribute multiple times so if you had a way to resolve the translation for an attribute you might do something like this.

def indexSolrTitle(doc) {
doc.addField("title_es_s), getSpanishTitle())
doc.addField("title_en_s), getEnglishTitle)
doc.addField("title_fr_s), getFrenchTitle)
}

What do you think? I need to think through the i18n use cases a bit more. Any suggestions for how you would like it to work?

Could solrAutoIndex be a global property? For instance ALWAYS reindex domain classes by default? And then maybe have a way to disable it globally (for bulk inserts)

The ability to hit the solr instance directly and test queries is really cool. I especially like the ability to see how many items are indexed.

Would be neat to wrap SolrQuery so that it could take a params map like .list() and .findBy() dynamic methods to facilitate pagination – so you could do
def query = new SolrQuery(params.q, params)
you could then look for offset and max in the params map and automatically do the pagination rather than the if statements you had in the controller

A solr paginate tag library that wraps the g.paginate would also be neat to have

Great work – looks really powerful! Can't wait to try it out

Sent from my iphone