I think long before IDEs implement any of these things - which represent a rather niche dev setup IMO - we need some way for libraries and applications to expos

I think long before IDEs implement any of these things - which represent a rather niche dev setup IMO - we need some way for libraries and applications to expose higher level metadata to language servers. Cross-repo search and replace is nice and all but it does't solve the problem currently filled by adhoc scripts and build steps that pull schema information from Swagger/Graphql/JSON schemas to generate types. Especially in the case of microservices where API boundaries are most often URLs or codegened RPC cliets, symbol renaming just isn't as useful outside of a single repo or workspace. Having, say, a protobuf or GRPC or OpenAPI language server feed information to the language server doing autocomplete and refactoring would be much more useful.