HoustonPutman commented on PR #16739: URL: https://github.com/apache/lucene/pull/16739#issuecomment-6297491146
Ok, I'm not an expert at all here, but this does look like a sane, good, and very concise change from a very birds-eye-view. Going through the original PR, and follow-ups, it doesn't look like this was an intentional choice, or at least there was no discussion on using graph size vs using live docs. However, there is maybe a better algorithm we could choose rather than just choosing the most vectors, such as taking into account deleted percentage to know whether or not we will go over the overall decay percentage that is allowed before doing some reconnection (added here: https://github.com/apache/lucene/pull/16559). But that's a bigger change, and this looks extremely well contained. @msokolov Do you have any thoughts here? I don't want this to hold up 10.6, and it looks like an easy commit and back port. But if you have concerns, we can easily hold it up for 10.6.1 (or a 10.7, if that ever happens). -- 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]
