Following up on TINKERPOP-2438 - I think I've addressed the performance issue that was noted. It's not a perfect solution but it was easy/obvious and I feel safe enough to merge:
https://github.com/apache/tinkerpop/pull/1368 I've formed the pull request for review with the intention to merge it for 3.4.9. Running docker tests now - please let me know if there are any concerns. On Mon, Nov 30, 2020 at 5:12 PM Stephen Mallette <spmalle...@gmail.com> wrote: > As we roll into 3.4.9 code freeze week we can use this thread for any > issues or concerns that are found related to testing this week. > > To get the ball rolling, a regression has been noticed in performance > around low-latency scripts and seems to be related to the new > GremlinASTChecker. My earlier tests on this feature didn't note the > cumulative damage of the additional AST processing that would occur on > small/fast scripts where it is much harder to hide millisecond range > increments. I've reverted that functionality: > > > https://github.com/apache/tinkerpop/commit/daad70bec96d4a1d3a0579bee1b73934d9293664 > > and reopened the issue: > > https://issues.apache.org/jira/browse/TINKERPOP-2438 > > I opted to not revert the entire body of work because I think the idea is > still valid in general but it needs more attention paid to these particular > queries. I also don't think it's a reason to pause 3.4.9 release over. This > release is too full of features to let this issue hold things up. > > I've published a 3.4.9-SNAPSHOT if you're like to test and I will post > back when freshy docs are published. > >