jimczi commented on PR #16214: URL: https://github.com/apache/lucene/pull/16214#issuecomment-4660728803
I wonder if this is the right tradeoff for ANN-backed indices in general. Losing the ANN structure for the whole segment after a small vector update and falling back to exact search until merge feels fairly expensive and potentially surprising. Maybe this only really makes sense for flat indices where there is no ANN structure to invalidate? The use case that feels more important to me is actually the opposite one: updating other fields/doc values without having to rewrite vectors or rebuild the ANN structure. I saw the “change of model” use case mentioned in the PR, but in that situation you typically need to reindex everything anyway, so I’m not sure this optimization helps much there. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
