Fail Drush Deploy if Database Updates Produce a Diff Against Tracked Configuration (#5389)

[adamzimmermann]

Motivation

• Fail Drush Deploy if Database Updates Produce a Diff Against Tracked Configuration #5712

Known Issues

• I targeted this against 11.x as the site I'm working on daily won't let me get to 12.x due to issues I can't resolve at the moment. I'm happy to port this to 12.x once we agree on the general approach. This just helps me test it more easily.
• Adding some better logging/visibility into the differences it detected would be great. I'm happy to add that too once we have an agreement on the general approach/messaging we want.
• I'm unsure if the return is the preferred way to abort a process here, but figured someone with more knowledge could quickly point me in the right direction.

Current State

I tested this locally and it does work though. I added an update hook that simply did this and it failed as expected. ✅

/**
 * Force a failure in the drush deploy command.
 */
function example_update_8004() {
  \Drupal::service('module_installer')->uninstall(['some_module']);
}

[weitzman]

Nice work! Also needs to go into 12.x first, and test coverage for should be expanded for deploy - that can be a followup.

[weitzman]

Could I get permission to push to your branch? Or you can accept my suggestions from the UI. This looks great to me. Lets address those Known Issues and then we can merge this.

[adamzimmermann]

I won't be on my computer much today, but believe I have the permissions adjusted. I'm happy to help with all of the needed adjustments next week, but I won't be around much this weekend. Hopefully I've unblocked things though. 🤞🏼

[markdorison]

Do we think this can be the default behavior? I would love it to be, but will this change be disruptive to some?

[weitzman]

I think it should only be disruptive when there is a genuine problem. Thus I think it runs always during a deploy. We can add a flag later if a need arises.

[adamzimmermann]

Thank you all for chiming in and pushing this forward. 🙏🏼

I'm hoping to find some time later today to target this to the 12.x branch and improve the logging.

Anything else that I should add to this before we consider is ready for a more final review?

I might need some help with test coverage too, but I'll take a swing at it.

[apotek]

Thank you all for chiming in and pushing this forward. 🙏🏼

I'm hoping to find some time later today to target this to the 12.x branch and improve the logging.

Anything else that I should add to this before we consider is ready for a more final review?

I might need some help with test coverage too, but I'll take a swing at it.

Thank you for your great work on this @adamzimmermann . The language (output) around this will be critical to a variety of users understanding why their deployment was blocked, and it strikes me that designing the user feedback will end up being a tangle in and of itself :)

[adamzimmermann]

Closing in favor of #5718.