While it is not a part of our daily CI, we are taking measures to detect Calcite's regressions early, including several preliminary integrations with Calcite's master between Calcite's releases. Let me once again highlight that this unfortunate situation with the late detection of CALCITE-3839 <https://issues.apache.org/jira/browse/CALCITE-3839> was caused by an unlucky timing of our new feature merge and not by a lack of CI.
On Tue, Mar 3, 2020 at 3:49 AM Francis Chuang <francischu...@apache.org> wrote: > On 3/03/2020 12:31 pm, Julian Hyde wrote: > > We can discuss the proposal to add other products' test suites to CI. > > I think it would cause more work for us (triaging false-negative > > problems). And lots of important downstream projects do not have a > > public test suite. > > Apologies for not being clearer, the comment regarding testing against > Calcite master in CI daily was directed towards Anton. > > It would not be practical for Calcite to run test suites for all of its > consumers. Consumers should be proactive in testing against our master > to detect regressions and breaking changes. > -- Best regards, Anton.
