Hi ufuk, thank you for sharing your insights. I am connecting sql handler through sqlalchemy (sqlalchemy_solr), a Python Library, and at this moment sql handler endpoint is the way to go. But yeah, it sounds promisse to resolve through streaming expressions.
Best, Koji Em seg., 21 de ago. de 2023 às 11:04, ufuk yılmaz <uyil...@vivaldi.net.invalid> escreveu: > > Oh I just tried and it seems possible to integrate sql handler and streaming > expressions, so it should be possible to aggregate sql results using stream > decorators in many ways: > > jdbc( > > connection="jdbc:solr://SOLR_ZK_CONNECTION_STRING?collection=COLLECTION_NAME", > sql="select NAME, ADDRESS, EMAIL, AGE from PEOPLE where AGE > 25 order by > AGE, NAME DESC", > sort="AGE asc, NAME desc" > ) > > Sent from Mail for Windows > > From: ufuk yılmaz > Sent: Monday, August 21, 2023 4:45 PM > To: users@solr.apache.org > Subject: RE: sql datetime field aggregation > > I didn’t use it too much but I’d guess syntax support for sql would be > limited to simple things. I’d love to learn if anybody has knowledge on the > sql handler. > > If your field is indexed using datetime field type or long number since > epoch, then range faceting could be a way to group based on month or year. > > Solr SQL uses streaming expressions under the hood, streaming expressions > offer a world of different abilities when querying, but you’d need to learn > it’s syntax. > > If you want to levarage your existing SQL knowledge on top of Solr, Apache > Spark could be an option. It can load data from Solr efficiently and run SQL > to analyze it, plus it supports nearly anything SQL can offer in terms of > aggregation. > > -ufuk yilmaz > > Sent from Mail for Windows > > From: Kojo > Sent: Monday, August 21, 2023 4:06 PM > To: users@solr.apache.org > Subject: sql datetime field aggregation > > Is there a way in Solr SQL to truncate data and aggregate it? for > example, by month or year in a datetime field? Or is the Solr approach > to create separate fields to enable these aggregations? > > Thanks in advance, > Koji > >
