fix ANTLR ATN version conflict between hibernate and cqengine

[lucastheisen]

• Brief description of changes applied
  • This fixes an ANTLR ATN verison conflict between upstream dependencies (hibernate and cqengine) by applying the upstream maintainers suggested workaround
• Test cases for all modified changes, where applicable
  • N/A, i could not come up with a small enough scenario
• The same pull request targeted at the master branch, if applicable
  • N/A, this PR is already targeted at master
• Any documentation on how to configure, test
  • N/A there is no documentation as it is a very minor build time change and to test, you would need to start up a cas that used both the google MFA and audit JPA and you got all the way through the authentication workflow to the point you got a response back from the google authenticator MFA (at least that is the point where this was failing from me and no longer does).
• Any possible limitations, side effects, etc
  • There are no limitations, and the only known side effect is a slighly larger dependency per the upstream maintainer of cqengine
• Reference any other pull requests that might be related
  • N/A this PR is a standalone

[mmoayyed]

Thank you for the pull request. Please put together a reproducer/test using the instructions provided in the PR template to demonstrate the problem.

[github-actions]

This pull request has been open and inactive. It will be closed shortly.

[lucastheisen]

@mmoayyed , sorry, i was on vacation last week...

i know this does not have "reproducer/test". That is because i cannot really conceive of a way to make that feasible. This issue arises when you complete a google auth MFA and have already loaded the hibernate JPA because hibernate JPA uses a new version of ANTLR that is incompatible with the version needed by the transitive dep on cqengine. I did provide links to the upstream (cqengine) library where the explicitly talk about this incompatibility...

is there no way to get a PR through where the issue is infeasible to try to force into a reproducible test?

[mmoayyed]

We have instructions and documentation provided in the PR template to help you understanding our testing process and how reproducers are put together. You're welcome to review.

is there no way to get a PR through where the issue is infeasible to try to force into a reproducible test?

Usually, no though this depends on the change. This is not infeasible. See ci/tests/puppeteer/scenarios/mfa-gauth-login-jpa