Open ml-evs opened 1 month ago
thanks @ml-evs, with basic indexes in mongodb, the default load can be made fairly fast except for the total number of structures (data_returned
). Might be worth adding this just for that number. Or alternatively use the mongodb estimated document count for the filterless query (but i guess this can be wrong is some very rare cases?)
Maybe this is a good idea after all, if a web client always hits the same request, we could cache these as JSON files that are served statically rather than via the API, e.g., via the config file:
"static_index": ["?page_limit=100&response_fields=id&page_offset=0"]
e.g., on server startup, take a series of chunked blobs corresponding to the default sort order of the API, then if a user requests with no filter and no page limit, return this static chunked data directly.
cc @JonathanSchmidt1 @eimrek (maybe this is a useful optimisation for e.g., the Li-ion set too)