fix: always rollback

[mitar]

Related Issue or Design Document

Fixes #637.

Checklist

• I have read the contributing guidelines
  and signed the CLA.
• I have referenced an issue containing the design document if my change
  introduces a new feature.
• I have read the security policy.
• I confirm that this pull request does not address a security
  vulnerability. If this pull request addresses a security vulnerability, I
  confirm that I got green light (please contact
  security@ory.sh) from the maintainers to push
  the changes.
• I have added tests that prove my fix is effective or that my feature
  works.
• I have added necessary documentation within the code base (if
  appropriate).

Further comments

There is not much change necessary to get this fixed. See the diff.

There is one logical change I have done: before rollback was not called if commit failed, now I made it that it is called if commit fails. It is generally safe to call rollback outside of a transaction in PostgreSQL (you only get warning), but if the connection is in error state, then rollback is required. So just to be safe, I think it is better to rollback on any error, including when commit fails (because we might not know what is the reason for failure of that call and in what state is the connection).

[aeneasr]

Are you still up for the changes? 🧐 If you need any help, let us know!

I'll convert this to a draft in the meanwhile. When it's ready for review again, please remove the draft indicator! 👀

[mitar]

Yes yes. I was just a bit busy with other things.

[aeneasr]

All good. I think I am known by now as one of the slowest maintainers in open source 😓

[mitar]

All good. I think I am known by now as one of the slowest maintainers in open source

What? No way. You are one of best maintainers I have seen around. Regularly checking in, always giving great feedback, pushing PRs through and not leaving them stick around.

[aeneasr]

What? No way. You are one of best maintainers I have seen around. Regularly checking in, always giving great feedback, pushing PRs through and not leaving them stick around.

You made my day, thank you :)

[aeneasr]

Re-running conformity tests that hung up

[mitar]

BTW, those tests are failing also in master branch.

[mitar]

ping @aeneasr

[aeneasr]

Ah yes it fails because the refresh PR has not yet been merged. I'd like for the OIDC conformity siute to pass first so that we can confirm that the rollback has no unintended effects :)

[mitar]

Which other PR?

[aeneasr]

i merged a workaround, let's see if we're green!