Fixes #36849 - Run GHA on Ruby 3.0

[ofedoren]

@ekohl, there is another, more general and detailed approach trying to fix the same issue. I hope the code and the comments will help to understand the idea of what's going wrong.

Unfortunately, that's the last thing I could find. The third and (that's only my opinion) the last option would be to completely get rid of that weird app/models/host.rb module, which (per my understanding) simply tries to simulate an abstract class.

UPD: TIL that rubocop can apparently yell that the indentation of a comment can be wrong and that's unacceptable 🤦

Alternative to #9871.

[ekohl]

I had a quick look at this and it appears the approach is to make the constructor's signature match the base class. That makes sense to me.

As for STI, I'm not sure we can get rid of it. I'd like to. At this moment I think we only have 2 implementations, so be sure to check out what Discovery does.

[ofedoren]

Even though initially I wanted to get the (any) fix merged at first, and then take a look at the plugins which fail, but after I've tested with foreman_discovery it seems some additional fixes might be needed in this/other PR as well, so I'll rather start a list here with verified plugins which support this PR:

• discovery: Move to inline settings validations foreman_discovery#622
• templates: Test on Ruby 3 foreman_templates#181
• webhooks: Test on Ruby 3 foreman_webhooks#72
• openscap: Test on Ruby 3 foreman_openscap#557
• tasks: Fixes #37103 - Ruby 3.0 support foreman-tasks#743
• REX: Test with ruby 3 foreman_remote_execution#864
• Katello: Fixes #37145 - Ruby 3 support Katello/katello#10885
• ansible: Test with Ruby 3 foreman_ansible#681
• google: Support Ruby 3.0 foreman_google#67
• puppet: Fix tests for Ruby 3.0 foreman_puppet#383
• bootdisk: Test on Ruby 3 foreman_bootdisk#156
• azure: Test on Ruby 3 foreman_azure_rm#184
• kubevirt: Test on Ruby 3 foreman_kubevirt#154
• leapp: Test on Ruby 3 foreman_leapp#136
• virt_who_configure: Test on Ruby 3 foreman_virt_who_configure#188

[ekohl]

Can we separate this out to its own commit? I'd like to merge all the preparation work before we pull the trigger.

[ekohl]

There are some deprecation warnings when running tests:

2024-01-23T18:54:00.9931233Z Mocha deprecation warning at /home/runner/work/foreman/foreman/app/services/proxy_api/dhcp.rb:36:in `unused_ip': Expectation defined at /home/runner/work/foreman/foreman/test/unit/proxy_api/dhcp_test.rb:99:in `block in <class:ProxyApiDhcpTest>' expected positional hash ({:query => "mac=00:11:22:33:44:55"}), but received keyword arguments (:query => "mac=00:11:22:33:44:55"). These will stop matching when strict keyword argument matching is enabled. See the documentation for Mocha::Configuration#strict_keyword_argument_matching=.
2024-01-23T18:54:01.0281310Z Mocha deprecation warning at /home/runner/work/foreman/foreman/app/services/proxy_api/dhcp.rb:36:in `unused_ip': Expectation defined at /home/runner/work/foreman/foreman/test/unit/proxy_api/dhcp_test.rb:74:in `block in <class:ProxyApiDhcpTest>' expected positional hash ({:query => ""}), but received keyword arguments (:query => ""). These will stop matching when strict keyword argument matching is enabled. See the documentation for Mocha::Configuration#strict_keyword_argument_matching=.
2024-01-23T18:54:01.0407752Z Mocha deprecation warning at /home/runner/work/foreman/foreman/app/services/proxy_api/dhcp.rb:36:in `unused_ip': Expectation defined at /home/runner/work/foreman/foreman/test/unit/proxy_api/dhcp_test.rb:90:in `block in <class:ProxyApiDhcpTest>' expected positional hash ({:query => "from=192.168.0.50&to=192.168.0.150"}), but received keyword arguments (:query => "from=192.168.0.50&to=192.168.0.150"). These will stop matching when strict keyword argument matching is enabled. See the documentation for Mocha::Configuration#strict_keyword_argument_matching=.
2024-01-23T18:54:13.7290895Z /home/runner/work/foreman/foreman/vendor/bundle/ruby/2.7.0/gems/fog-vsphere-3.6.3/lib/fog/vsphere/compute.rb:344: warning: deprecated Object#=~ is called on Integer; it always returns nil
2024-01-23T18:54:13.7458370Z /home/runner/work/foreman/foreman/vendor/bundle/ruby/2.7.0/gems/fog-vsphere-3.6.3/lib/fog/vsphere/compute.rb:344: warning: deprecated Object#=~ is called on Integer; it always returns nil
2024-01-23T18:54:22.8953091Z Mocha deprecation warning at /home/runner/work/foreman/foreman/app/models/host/managed.rb:715:in `ipmi_boot': Expectation defined at /home/runner/work/foreman/foreman/test/models/host_test.rb:3245:in `block (2 levels) in <class:HostTest>' expected keyword arguments (:function => "bootdevice", :device => "bios"), but received positional hash ({:function => "bootdevice", :device => "bios"}). These will stop matching when strict keyword argument matching is enabled. See the documentation for Mocha::Configuration#strict_keyword_argument_matching=.

The fog-vsphere warnings will break with Ruby 3.2. The other ones look like some things we should fix in our code and then enable strict_keyword_argument_matching in mocha.

[ekohl]

Can we also enable strict keyword argument matching (https://mocha.jamesmead.org/Mocha/Configuration.html#strict_keyword_argument_matching=-instance_method)? Perhaps make that the whole second commit?

[ofedoren]

Done. Not sure though if we need that. As there was stated, this will be true by default anyway.

[evgeni]

Done. Not sure though if we need that. As there was stated, this will be true by default anyway.

That way we don't accidentally regress now that we fixed the offenders until it's true in the future?

[ekohl]

Indeed. If it was true today then these tests wouldn't have failed. This way we also force plugins to adopt.

[ofedoren]

Some Katello test failures are related to strict keyword argument matching I guess, but otherwise 🍏 for Foreman.

[ekohl]

Some Katello test failures are related to strict keyword argument matching I guess, but otherwise 🍏 for Foreman.

I've love to leverage what you did in #9989 (comment) and see how other repositories behave.

[ofedoren]

I've love to leverage what you did in #9989 (comment) and see how other repositories behave.

I'm preparing more PRs regarding what I found in Katello, I'll paste them here once they are ready. The plan to go over my locally checkedout other plugins as well.