mirror of
https://github.com/qdrant/qdrant.git
synced 2026-08-05 01:20:53 -05:00
prompt: These recent changes precollect all bytes that can be used during the query before actually running the query. This is fine, and desired, but the PR takes it a step further and precollects the evaluation of the data. This can potentially be expensive since we need to process the entirety of the data. It is done like this so that lifetime tracking gets easier. But I have a challenge for you. Can you make it so that the evaluation is preserved lazily? Meaning, to keep using iterators as outputs.