Default arguments if `fails`

[youroff]

Not sure if that's a problem of CoffeeScript which we use, or just something general, but when we use something like:

 mockUpdate(@user).match(first_name: '').fails(status: 422, response: {errors: {first_name: ["can't be blank"]}})

It takes convertErrors as false. So I guess it's some issue with the args notation here:
https://github.com/danielspaniel/ember-data-factory-guy/blob/master/addon/mocks/mock-request.js#L43
Might worth rewriting it good 'ol way to support CoffeeScript users.

[youroff]

Though it's weird that it doesn't work for me, cause CoffeeScript does support destructuring assignment and after compilation it must be just as passing plain object into a function with such arguments. Maybe that's babel that takes that stuff off on transpiling?

[danielspaniel]

hm .. that is kind of a bummer because me like that es6 notation alot.
not sure why would make a difference if using coffescript .. but I will write something in coffeescript and see if I get same problem and then undo that fancy notation if I have too :( will make me sad though.
Can you not use coffeescript ( ?? ) ..
Did I just say that?
I took all my tests in this one project and started converting away from coffee a few months ago, since es6 just about makes it not necessary anyway .. and I feel like it is the right way to go anyway .. so .. just throwing that out there.
but I agree it should not break for coffeescript users

[youroff]

Well, you're the owner of this lib, so that's your call :) I can explicitly set properties in such cases, which actually I did when figured out the problem, but other users can run into this issue as well, that's my main concern. So it can be told somewhere in documentation, as an easiest option.

I like that shiny ES6 features too, and if they were there when I started using Coffee, may be I wouldn't :) And to be honest, that should work just fine, as exactly the same notation of destructuring arguments assignment was in CS before, so maybe it's something tiny that could be fixed easily.

[danielspaniel]

I agree though about everything. It should work for coffeescript users .. I am sure that is a bug somewhere in the chain of happenings ( babel to coffee compilers ) .. so, yes I will fix it ..

[youroff]

We can help with that if you want. Also I have couple of ideas, for one of which my mate is going to make a PR by tomorrow.

[danielspaniel]

Now your talking bro .. I'll take the help .. I am a bit overstretched these days .. and this kind of fix is pretty swift and painless

[youroff]

Deal. Will prepare a PR soon

[danielspaniel]

how are things going with the pr .. need any help ?

[youroff]

Sorry, been pretty busy. Will try to get to this soon!

Thanks for #231 btw! 👍 👍 👍

[youroff]

I think I need your comment on this one:
https://github.com/danielspaniel/ember-data-factory-guy/blob/master/addon/mocks/mock-update-request.js#L54

Who said that update errors don't need conversion? :-\ Now I'm puzzled too... I don't understand the reason of overriding this method.

[danielspaniel]

I think I ran the tests and I needed to have that override ( which puzzled me ). But if things now work without it .. be my guess and blow that away.

[youroff]

So that means there's somewhere in the tests wrong expectation? Anyways, lemme check that.

[youroff]

Oh, I see... So to fix that we either have to pass convertErrors: false in tests, or expect errors in appropriate format:
https://github.com/danielspaniel/ember-data-factory-guy/blob/master/tests/unit/shared-factory-guy-test-helper-tests.js#L1732-L1751

But there's another problem, I thought that convertErrors option is Adapter specific, but when true it always returns json-api style errors. Should we delegate conversion to adaptors instead?

[danielspaniel]

The errors end up json-api style in the end anyway ( as I recall )
If that works ( to delegate to adapter ) .. then great, so I don't mind either way.