On 24/08/2024 19:14, Dmitri Maziuk wrote: > > If the application specified the number of shards and replicas when it > created the collection, it should know what it asked for.
This is a separate issue, but purely from the perspective of an application developer, I don't feel like I should have to specify that information at all: That seems like something that should be up to the administrator of the solr server(s). > While I agree that an easy button for uploading a complete schema.xml > could be convenient in some cases, I don't know if it's that simple to > implement... whereas writing a script to transform schema.xml into a > series of curl commands, is. Unless I'm missing something obvious, it doesn't seem all that simple. The reason being: Theoretically, I don't know what the contents of the initial schema are, because I'm an application being installed next to existing solr servers, and their default configuration could be anything. Is there, perhaps, an obvious way to completely zero out a schema to nothing so that I can add types and fields and be confident that only the types and fields that I explicitly added exist? -- Mark Raynsford | https://www.io7m.com
