Throw error on undefined value from reducer function

[taylorhakes]

Accidentally created against alpha before. Sorry about that.

Updated composeReducers to throw error when reducer returns undefined. Update to #193. #191

[taylorhakes]

@gaearon Is this good to merge?

[gaearon]

Yeah, can you please rebase so it merges cleanly?

[taylorhakes]

Just did the rebase. The tests pass locally, but are failing in Travis. Have you seen this before?

[gaearon]

Something to do with linting. I'll merge soon, thanks.

[acdlite]

Sorry for only now bringing this up, but I believe the check should go inside the Store class rather than composeReducers(). That way it works for all reducers, not just those created by that function. Also, we use the invariant module for consistency, like @gaearon does here: https://github.com/gaearon/redux/blob/breaking-changes-1.0/src/Store.js#L26-L29

[acdlite]

^ In fact the new invariant should go just a few lines below the highlighted portion

[gaearon]

Sorry for only now bringing this up, but I believe the check should go inside the Store class rather than composeReducers().

But then we'll check the root value (which is unlikely to be undefined if you use composeReducers).

[gaearon]

I think we're actually doing the right thing here. First, we want to check the actual app's reducers—not the root one. Second, we don't want this to be a hard limitation—just something the default utilities help you with. You're free to use your own composeReducers that allows undefined for individual reducers, or even not use composeReducers at all.

[acdlite]

@gaearon Ah I see. It should go in both places, then, right?

[acdlite]

Okay I can agree with that. That makes sense.

[gaearon]

Why do that though? Redux doesn't really care what your root value is like. I'd say it should only be enforced at composeReducers level, as our built-in utilities follow “make it easier for a beginner” approach.

[acdlite]

Yep, agreed. Carry on :)

[taylorhakes]

Updated code to use invariant

[gaearon]

Can you please rebase on master again? I think the linting problem is fixed there by your earlier PR.

[gaearon]

missing newline before it

[taylorhakes]

I am a little confused. This change was made on top of breaking-changes-1.0. Do you want want me to merge master into breaking-changes-1.0? This change can't be rebased on master because it has to do with reducers.

[gaearon]

Oh, I didn't mean on master, sorry :-(
I just updated breaking changes branch.

[gaearon]

All right, this is in.
I slightly tweaked the tests and error messages: de685de.