[
https://issues.apache.org/jira/browse/DRILL-8412?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17702884#comment-17702884
]
ASF GitHub Bot commented on DRILL-8412:
---------------------------------------
jnturton commented on PR #2776:
URL: https://github.com/apache/drill/pull/2776#issuecomment-1476733563
> @jnturton I think we should definitely run this experiment. I'm also
curious as to how to run this with a specific PR from Calcite. That way I can
contribute to the review process over there and find things that break drill
quickly.
They won't build and publish artefacts for unmerged PRs (would be a security
problem) so we'll have to run our own Calcite builds for these. Probably the
best is for the developer to pull the PR into a local branch and build Calcite
putting the results in their local Maven repo, then build Drill.
> Upgrade to Calcite 1.35-SNAPSHOT
> --------------------------------
>
> Key: DRILL-8412
> URL: https://issues.apache.org/jira/browse/DRILL-8412
> Project: Apache Drill
> Issue Type: Task
> Components: SQL Parser
> Reporter: James Turton
> Assignee: James Turton
> Priority: Major
>
> This issue proposes that we try basing Drill master on snapshot builds of the
> upcoming version of Calcite so that the CI tests that run automatically upon
> commits to master will exercise Drill with present day Calcite.
> 1. The CI tests that run automatically upon commits to Drill master will
> exercise Drill with present day Calcite.
> 2. Breaking changes in Calcite would (mostly) break the Drill CI and force us
> to deal with them in order to proceed.
> 3. Regressions in Calcite would (mostly) break the Drill CI and force to
> report them in order to proceed.
> 4. If Drill master becomes too unstable when it is based on Calcite snapshots
> then this change is trivially undoable.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
