I’ve fixed SOLR-11313 and SOLR-11315. I also found a couple doc and source comment trie* field mentions that I got rid of: SOLR-11324.
-- Steve www.lucidworks.com > On Sep 2, 2017, at 6:17 PM, Jan Høydahl <jan....@cominvent.com> wrote: > > Found a related blocker: > > https://issues.apache.org/jira/browse/SOLR-11315 "Files example does not > start due to "Unknown fieldType 'tdouble’"" > > The fix for both DIH and files example seems to be just replacing “t” with > “p” in the schema. I just tested, but got some other WARN logs as well, see > JIRA. > Will leave the fix to someone in a more convenient time zone :) > > -- > Jan Høydahl, search solution architect > Cominvent AS - www.cominvent.com > >> 2. sep. 2017 kl. 23.47 skrev Jan Høydahl <jan....@cominvent.com>: >> >> Thanks Alexandre for doing some extended tests instead of just running the >> smoke tester. >> There should be unit tests that validates that all the examples spins up >> without errors… >> >> Created https://issues.apache.org/jira/browse/SOLR-11313 (7.0, Blocker) >> >>>> Additional comments that probably cross several JIRAs: >>>> *) The schemaless mode duplicates auto-guessed text fields to string >>>> fields. Those string fields are docValues=true, indexed=false, >>>> stored=false. But docValues do return their string representation so >>>> the query now returns the same content in both guessed text field and >>>> copied string field. I am not sure if that was the intention. >>> >>> useDocValuesAsStored="false" is needed on those fields perhaps? >> >> At the other hand, seeing the fields explicitly in the response will also >> remind the user that >> they are there and can be used. We recommend auto-guessing for getting >> started, and then >> ask users to turn it off before production and tune the schema. Flipping the >> useDocValuesAsStored >> bit could be one such advise, not because it will save any disk space, but >> to get rid of the >> redundant output. >> >>>> *) In README.txt we tell how to start a server in 3 different ways, >>>> but not at all how to stop one. Should we? >>> >>> Definitely! >> >> https://issues.apache.org/jira/browse/SOLR-11312 (7.1, Trivial) >> >>>> *) http://lucene.apache.org/solr/quickstart.html is now out of date in >>>> couple of ways apart from the version >>>> **) for the indexing of "docs/" as there is only 1 indexable file in >>>> there are Javadocs were removed >> >> It’s been discussed in some documentation JIRA, but I agree that it should >> be fixed >> when 7.0 is released. >> >>> >>>> *) techproducts example starts, but returns fileds like >>>> "price_c____l_ns":0 , which I am not sure what is supposed to >>>> demonstrate/represent. >>> >>> Ugh.. that is a pseudo-field being returned for the currency field >>> type (not sure when that copyfield was added). >>> Presumably the pseudo-field is being returned because it has docValues >>> (which we can treat as column-stored fields now). >>> Any copyField targets, as well as pseudo-fields, should have >>> useDocValuesAsStored="false" set. >> >> Sounds like a good prescription. I have not created a JIRA for this... >> >> -- >> Jan Høydahl, search solution architect >> Cominvent AS - www.cominvent.com >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
