pauldoo commented on PR #16145: URL: https://github.com/apache/lucene/pull/16145#issuecomment-6792086210
> Is the only remaining debate what to default SKIP to? How do we settle that so we can merge? Thoughts? **TLDR:** I still have a preference for a larger default (maybe 256 that I think everyone looks happy with?), but I wouldn't be upset if the proposed code was merged as-is with the value of 64. **Longer version:** Today in main the check interval is effectively infinite on a warm input. Moving from infinity to 64 is more aggressive than from infinity to 128 or 256. This alone makes me prefer a larger value where we're more confident the periodic checks will not cause a noticeable regression in some workload. Could we merge it with a value of 256, which I think everyone was happy with? https://github.com/apache/lucene/pull/16145#issuecomment-6462845779 > With this analysis, perhaps 256 is a reasonable middle ground given the various comments that have mentioned that value? https://github.com/apache/lucene/pull/16145#issuecomment-6376381113 : > So I'd go with 128 as a middle ground but 256 is fine with me too. This would allow the PR to be merged delivering the bulk of the value. We can then have a followup discussion on lowering the default to be more aggressive or using `perf_event_open` / `getrusage`. --- Having said all that, I do accept that the risk of regression looks small with a default value of 64. If Lucene maintainers wish to merge the currently proposed code I'm happy with that too. -- 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]
