Update API to use descendant_get to return valid types

[jntullo]

Mixin that returns the correct subclass from the specified string, or raises and error if it is not a subclass of that type.

Part of a refactoring effort for #14217

@miq-bot assign @Fryguy
@miq-bot add_label refactoring, wip

[jntullo]

@Fryguy wasn't sure we should default to returning itself if type is nil. We did this in Authentication, not sure we should generalize that though. thoughts?

[Fryguy]

I don't like the class_from_string method as it seems to imply you can get any class, but you really can't. Perhaps, descendant_by_name or descendant_for? This seems to imply you will get the descendant class, and it mirrors the method named descendants

[jrafanie]

haha, I didn't see this until now. Honest.

[miq-bot]

This pull request is not mergeable. Please rebase and repush.

[chessbyte]

@Fryguy bump

[Fryguy]

@jntullo I commented above, but also was thiking that maybe this can be part of more_core_extensions, since it's so generic. @bdunne @jrafanie Thoughts?

[jrafanie]

I had to look at the tests to know what class_from_string did.

descendent_class_from_name

Naming is hard. Technically, it should be descendent or self but that's just too many words.

I like the idea, just not sure about the name.

[jrafanie]

@jntullo I commented above, but also was thiking that maybe this can be part of more_core_extensions, since it's so generic. @bdunne @jrafanie Thoughts?

Maybe? I mean, it makes sense to go there because it's a plain ruby enhancement.

[jntullo]

@Fryguy @jrafanie naming really is hard. I will update the name, and yeah, I think it might make sense for more_core_extensions as well. I'll put something together

[Fryguy]

Summary of names:

descendant_by_name
descendant_for
descendant_class_from_name
descendant_from_name

I think I like the last one the best, personally. It takes @jrafanie's but removes the _class bit since it's not needed, just as the descendants method isn't called descendant_classes.

[Fryguy]

Technically, it should be descendent or self but that's just too many words.

The ancestry gem differentiates this with the word subtree, where subtree is descendants + self: https://github.com/stefankroes/ancestry#navigating-your-tree . Even so, I can't think how you could write a method name with subtree in it and still make sense. subtree_from_name isn't getting the whole tree, so it makes no sense. 😕

[bdunne]

👍 to descendant_from_name

or descendant_named
or descendant_const_get

[jrafanie]

I didn't look, are there no other callers? Can you grep not only manageiq but all providers to make sure we update all callers?

I usually do this now that we have so many extracted repos..

$ bundle show rails
/Users/joerafaniello/.gem/ruby/2.4.1/gems/rails-5.0.4

Take that path and basically grep the grandparent directory so it includes not only gems but bundler git gems:

grep -ir "class_from_request_data" /Users/joerafaniello/.gem/ruby/2.4.1

Or something like that

[jntullo]

@jrafanie yeah this is the only caller - this was a method I created for the API back in March to be replaced by this update

[jrafanie]

Nice, I like it.

[jntullo]

will need new version of more_core_extensions to complete this PR

[Fryguy]

There's one other PR in the works on more_core...just want to get that one over the finish line as well before cutting a release. cc @bdunne

[bdunne]

@jntullo Can you update the minimum version of more_core_extensions in the Gemfile?

[miq-bot]

Checked commits jntullo/manageiq@3ba7f66~...25d4e5b with ruby 2.2.6, rubocop 0.47.1, and haml-lint 0.20.0
6 files checked, 0 offenses detected
Everything looks fine. 🍰