[
https://issues.apache.org/jira/browse/OFBIZ-11304?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17136784#comment-17136784 ]
Jacques Le Roux commented on OFBIZ-11304:
-----------------------------------------
OK, [I had a discussion with Daniel Gruno on Infra Slack Channel|
https://issues.apache.org/jira/browse/INFRA-20314?focusedCommentId=17135759&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17135759] and clearly they can't install a git checkstyle pre receive hook as requested in INFRA-20314. I have to check:
https://github.com/isaacs/github/issues/1002https://github.community/t/pull-request-action-does-not-run-on-merge/16092Here are my thoughts for now.
h6. Drawbacks with our current solution:
# We can't prevent PR merges to contain style issues.
# It consumes CPU locally
# It forces users to use it OOTB (but they can bypass or remove by hand easily, see below)
# What about checking branches? The hook is present in all branches but not checkstyle
h6. Advantages of our current solution:
# Check the compilation, tests (not integrations tests) and style, before pushing
# Can be bypassed by using --no-verify option or removing the hook by hand in build.gradle
> Install a Checkstyle pre-push (on every committer machine?)
> -----------------------------------------------------------
>
> Key: OFBIZ-11304
> URL:
https://issues.apache.org/jira/browse/OFBIZ-11304> Project: OFBiz
> Issue Type: Sub-task
> Affects Versions: Trunk
> Reporter: Jacques Le Roux
> Assignee: Jacques Le Roux
> Priority: Minor
> Fix For: Upcoming Branch
>
>
> The ofbizTrunkFrameworkPlugins build fails when a lint error is detected by the check gradle task. It's "hard" to exactly know from where lint errors come among all still present.
> I think we should rely on a Checkstyle pre-commit hook like
https://gist.github.com/davetron5000/37350 to complement tasks.checkstyleMain.maxErrors. This pre-commit hook prevents to commit when a lint error is present in the commit.
> Every committer would have it installed locally and the problem would be gone with some committers good will. I started a discussion about it at
https://markmail.org/message/guxbsvdkzky7gtdx. Jacopo made the same proposition years ago:
https://markmail.org/message/gkgmko4axj3vtnv3--
This message was sent by Atlassian Jira
(v8.3.4#803005)