tdunning commented on pull request #2432: URL: https://github.com/apache/drill/pull/2432#issuecomment-1018676376
The meaning of the compression parameter became a bit more strict in this latest version. As such, the digest should be a bit smaller and tails should be a bit more accurate. It is OK to increase the compression parameter if you want, but it has to be recognized that the results are approximate and a bit randomized. Can you describe a bit more about what the test is looking at? I could comment more specifically about what to expect if you like. On Fri, Jan 21, 2022 at 3:55 AM James Turton ***@***.***> wrote: > Hi @tdunning <https://github.com/tdunning>. This PR includes an update of > t-digest from 3.2 to 3.3 and I think that has caused a unit test of > metastore statistics to fail with > > Error: Failures: > Error: TestAnalyze.testHistogramWithDataTypes1:501 Did not find expected pattern in plan: Filter\(condition.*\).*rowcount = 676.*,.* > > , where what's changed is the row count estimate for the results of an > integer inequality predicate store_id < 15 which has gone from 676 to > 699. Have some approximate stats computed by t-digest shifted in this > version, e.g. an accuracy improvement? If so, and this is expected, then > I'll update the unit test's expectations accordingly... > > — > Reply to this email directly, view it on GitHub > <https://github.com/apache/drill/pull/2432#issuecomment-1018440761>, or > unsubscribe > <https://github.com/notifications/unsubscribe-auth/AAB5E6SSUDJZ52QF6OIMLK3UXFCTLANCNFSM5MM2LDFQ> > . > Triage notifications on the go with GitHub Mobile for iOS > <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> > or Android > <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. > > You are receiving this because you were mentioned.Message ID: > ***@***.***> > -- 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]
