Todd Lipcon has posted comments on this change. ( http://gerrit.cloudera.org:8080/9440 )
Change subject: KUDU-2312: Scan predicate application ordering is non-deterministic ...................................................................... Patch Set 4: Code-Review+2 > No, we put the predicates into an hash map, and at that point they lose all > ordering (ScanSpec::predicates_). The /scans page is also populated from > this unorderd map, so it doesn't reflect the true ordering either. I don't > think it's wise to guarantee an ordering externally. Ideally we'd be using > our own internal stats to do the predicate evaluation ordering on a per-cfile > basis, so there wouldn't even be a consistent global ordering. OK. Well, you've convinced me that this patch isn't actively harmful. Given we don't have any internal cardinality stats, though, I think our next step should be to allow Impala to pass through its own ordering explicitly when it _does_ have stats. We should see a good speed boost from that until we get to eventually computing our own cardinality estimates per-DRS or per-tablet. -- To view, visit http://gerrit.cloudera.org:8080/9440 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I99b2cabecd8626cad7e11fbdd492af7276e08348 Gerrit-Change-Number: 9440 Gerrit-PatchSet: 4 Gerrit-Owner: Dan Burkert <danburk...@apache.org> Gerrit-Reviewer: Dan Burkert <danburk...@apache.org> Gerrit-Reviewer: Kudu Jenkins Gerrit-Reviewer: Thomas Tauber-Marshall <tmarsh...@cloudera.com> Gerrit-Reviewer: Tidy Bot Gerrit-Reviewer: Todd Lipcon <t...@apache.org> Gerrit-Comment-Date: Tue, 06 Mar 2018 18:50:46 +0000 Gerrit-HasComments: No
