It's a common problem with pure negative query. The most problematic thing is that it's fixed in %50 cases, keeping users unaware of it. Let's track it if someone wants to kick tires https://issues.apache.org/jira/browse/SOLR-16247.
On Thu, Jun 9, 2022 at 8:45 PM Thomas Woodard <[email protected]> wrote: > Yes, thank you! That produces results. > > On Thu, Jun 9, 2022 at 12:04 PM Mikhail Khludnev <[email protected]> wrote: > > > Hi, Thomas. > > What if you try childFq=*:* -sku_type_s:25 ? > > > > > > On Thu, Jun 9, 2022 at 7:16 PM Thomas Woodard <[email protected]> > wrote: > > > > > This is in 8.11.1. I've got the following (parsed) query that returns > no > > > results: > > > { > > > "q":"{!parent tag=top filters=$childFq which=scope_s:parent > > > v=$childQuery}", > > > "json.facet":"{colors:{ domain: { excludeTags:top, > > > filter:[\"{!filters param=$childFq excludeTags=color v=$childQuery}\", > > > \"{!child of=scope_s:parent filters=$fq}scope_s:parent\"] }, > type:terms, > > > field:color_ids_ss, limit:-1, facet:{ > > parentsCount:\"uniqueBlock(_root_)\" > > > } }}", > > > "fl":"id", > > > "edisQ":"XXXXX", > > > "fq":["available_catalog140002_b:true", > > > "{!edismax tag=top qf=\"name_t^6 name_s^10 studio_t^2 series_t > > > categories_catalog140002_txt ancestor_categories_catalog140002_txt > > > features_txt selling_points_txt\" pf=\"name_t^6 studio_t^2 series_t^2 > > > categories_catalog140002_txt^2 ancestor_categories_catalog140002_txt^2 > > > features_txt selling_points_txt\" mm=\"2<-1 4<60%\" tie=\"0.1\" > > > v=$edisQ}"], > > > "rows":"28", > > > "childFq":["available_catalog140002_b:true", > > > "-sku_type_s:25"], > > > "childQuery":"scope_s:child"} > > > } > > > > > > If I take out childFq=-sku_type_s:25, it works. If I change it to > > > childFq=sku_type_s:21, it works. I tried integer fields originally, and > > > thought that maybe the problem was not being able to negate a filter on > > an > > > integer, but changing to string did not fix it. I'm stumped finding an > > > explanation for this, but I have also found no examples online that > > combine > > > real complexity with block join query parsing. Any suggestions on what > I > > am > > > doing wrong? > > > > > > > > > -- > > Sincerely yours > > Mikhail Khludnev > > > -- Sincerely yours Mikhail Khludnev
