is it required that :keyfn be an integer? When prerendering an om.next app, when trying to pass to :keyfn a fn that returns either a keyword or a string, the compiler complains: java.lang.IllegalArgumentException: Key must be integer

oops, scratch that

I’m being stupid

typo 😛

Does every entry in the om next query cause call to read with the entry as the key? Eg I want to have UserList and a User, what's the best way to fetch the data from remote? Define :user/list read key? Is there a definitive guide on the query expressions and read, or is it mostly blogs and wiki pages?

@alex-glv Generally, only the individual queries of the root component will trigger reads in the parser.

(defui ChildComponent
  static om/IQuery
  (query [this]
    [:child-query1 :child-query2])) ;; these don’t trigger reads

(defui RootComponent
  static om/IQuery
  (query [this]
    [:query1 :query2 {:join-query (om/get-query ChildComponent)}])) ;; Generally, compose all subqueries into root query.
;; triggered reads are: :query1 :query2 :join-query

I would place the remote query in the :user/list read key.

I don’t think we’ll have a definitive guide until om.next settles down into not-alpha. In the meantime, this is a very good resource: https://anmonteiro.com/2016/01/om-next-query-syntax/

Awesome, thanks!

👍

Is it ok to have ident’s on the components in a union?

@gardnervickers: I think it's a requirement for a union query

Anyone have any experience writing Om components and using them within a JavaScript project?

The best strategy I have at the moment is to have the JavaScript project create a new div as an injection point for an Om root. This would involve have a React root (built with Om) inside of another React root (built by the top-level JavaScript project).

what approaches have people been using for mocking the parser in testing the ui? I'd like to do some browser-based testing and want to test the parser logic, but don't want to set up a backend to handle remotes.

@grzm you don’t need to setup backends

the parser never sends anything to the remote

the result of parsing something in “remote mode” is a query expression

not a remote call

So when I create a reconciler, I just leave the remotes empty and it should all be good?

depends on what you want to test

lol. At this point I want to make sure I haven't broken mutations when I do refactoring/code updates. Like when I click a button that triggers a transaction I'm getting the appropriate state change.

what I meant is: you can test remote interactions, and if you want to do that, you don’t want to leave remotes empty in the reconciler

So I guess there are two things: one is that the components are rendering properly, and that the mutations work

you can e.g. have a different send function

Yeah, right now I'm not worried so much about full-stack remote interactions.

Okay, so it sounds like I should be looking at send

Cheers!

@grzm so one thing you might have not realized

There's a lot I have not realized 🙂

is that you don’t need to assemble a reconciler

you can test that mutations work with only a parser

(def p (om/parser {:read read :mutate mutate}))
(p {:state (atom init-state)} '[(do/this!)] )

And the result value is the updated state?

well...

since the state is an atom you can have a reference to it somewhere and deref it for checking the result, right?

Gotcha

I don’t remember what the result of a mutation is from the top of my head

you’ll have to check it out for yourself 🙂

You sure? I mean, you can just shell into my machine and do my work for me 🙂

@grzm sure, I’ll let you know my hourly rate 🙂

I'm confused about this error, (str "No queries exist for component path " cp " or data path " path') https://github.com/omcljs/om/blob/3a8229c4a843c9978ced8eec94f0fa33ae57b869/src/main/om/next.cljc#L1962

Shouldn't that be "and data path path'"?

That error only comes up if at least one query is found for the class path, but none of them matches the data path.