The closest things I could find to a solr implementation of saved search are:
1. solcolator https://github.com/SOLR4189/solcolator 2. Solr-monitor (unrelated to the proposed saved-search module) https://github.com/o19s/solr-monitor The problems with these is they treat the lucene-monitor (originally luwak contributed by OSC) data structure as a side-car and never intend to have solr manage the data so you miss out on consistency, replication, etc. The solution I propose is to treat saved-search as a first class citizen because, if you squint, it really isn't _that_ different from regular document search. I've chatted with Alan on this lucene PR https://github.com/apache/lucene/pull/13993 which was motivated by the latest proposed saved search module. Perhaps Alan, if they are still on the solr list, can chime in about any such work they did. It's possible there is some module out there that I just couldn't find that can be dusted off. I do know Charlie Hull is on this list might have a better idea about the genesis/pedigree of lucene-monitor (luwak) as well as any possible solr integrations that may exist that obviate the proposed saved-search module :-) I know that Solr's never-released-past-beta topic streams were related to this but I don't think it ever aspired to be a performant reverse search solution. It's theoretically possible to merge the two efforts although I think supporting fast, percolator-like reverse search via a request-response (or response stream) API is worthy enough to stand on its own. Sent from Bloomberg Professional for Android ----- Original Message ----- From: David Smiley <users@solr.apache.org> To: users@solr.apache.org At: 12/21/24 15:27:55 UTC-05:00 A huge feature! Can you please remind me of the historical pedigree of the underlying functionality? You point to "lucene-monitor" but that in turn came from somewhere. If I'm not mistaken, there was (is?) a Solr plugin out there from which this started, built by Alan Woodward. On Sun, Dec 8, 2024 at 4:47 PM Luke Kot-Zaniewski (BLOOMBERG/ 919 3RD A) < lkotzanie...@bloomberg.net> wrote: > Hello, > > Have you ever wondered if Solr will ever support saving searches and > efficiently > surfacing them given a matching document? > > I can't answer that question with certainty but still wanted to share a > bit of work > that has been done over the last year to that end. The (proposed) > saved-search module > attempts to provide a request-response API which takes a document and > fetches matching > queries, something similar to Elastic's Percolator. Solr's streaming > expressions API > should work with this as well if you need to fetch a very large number of > queries. > > The documentation is here (still a WiP) > > https://github.com/kotman12/solr/tree/solr-monitor-doc/solr/modules/saved-search#overview > > If you want to try it out yourself you can run with the patch attached to > the latest update > on the JIRA: > https://issues.apache.org/jira/browse/SOLR-4587 > > This might be of special interest to you if you already use lucene-monitor > for this > functionality. The proposed module uses all the clever lucene-monitor > optimizations > for document <-> query translations. On top of that, it comes with a few > of its own > optimizations around caching which result in potentially lower memory > utilization as > well as freeing queries from getting blocked by updates, which is > something you may > have encountered running lucene-monitor on your own. It also scales by > default (thanks, Solr) > and lets you seamlessly reuse your Solr schema + analyzers for > saved-search. > > I would greatly appreciate feedback or involvement from anyone on this > list who might > be interested. Looking forward to hearing from you here or on the PR .. or > the JIRA. > > Thanks, > Luke
