I don’t like defining new classes with Struct.new( some_symbols), but I do occasionally use OpenStruct as the base for a model. Your AwesomeResource seems like a nice compromise between the two: your attributes are dynamic, but accessing non-existent ones raises an error.

Then I started to poke around in irb until I came up with this:

https://gist.github.com/russen/5645979

The ‘if/else’ isn’t as OOP as your approach, but since the function is only 4 lines, it does pass Sandi Metz’s 4 rules :) (https://gist.github.com/henrik/4509394)

I had fun trying out this different solution, and am completely interested in any +/- feedback you have!

When you call attr_* methods on class, you define methods for all instances of that class, not just the current instance. You’d probably want to use `define_method` if you go that route. The main reason I didn’t use `define_method` is because once you save a resource, the server responds with the new authoritative attributes for that resource. If those attributes are different from what you sent, you have to undefine the methods. Not the end of the world, but it’s some manual work that I’ve decided to avoid for now.

-Parker

Thanks for the question. The main reason I didn’t explicitly define methods is because after a save, the response that comes back from the server is the new authoritative version of that resource, and if the attributes are different from what you sent, you’ll have to go back and undefine the old attributes and define new attributes.

Thanks!
-Parker

Also as you called out, mutual respect is crucial through every step of the process. An abundance of mutual respect is one of the reasons I love working at Pivotal Labs. Products are always better when the designers and developers realize that they’re all on the same team.

Thanks for sharing! Looking forward to reading more of your posts.