I think you got something wrong here. Fulcro is a client library.

I guess you're right, I don't see the full picture. Does not fulcro, if you use the http remote for networking by default use transit to send EQL queries to the server where the resolvers for said queries are situated?

Fulcro is definitely not just a client library. It’s fullstack. But I wouldn’t use it for scenarios where you don’t have full control over the clients. You could separate out your business logic from your resolvers/mutations and then simply add an api for external clients. I think fulcro is amazing for any client you control (Even mobile. I’m using fulcro and react native to do that). It allows for much faster iteration cycles. But I think it’s generally advisable to keep a separate API for clients that you don’t control, and simply have it call into your business logic as needed. • Fulcro apps (mobile, web, desktop) • Graphql or REST api or none EQL client • Business logic (The two above calls into this part) You could use https://polylith.gitbook.io/polylith/ or just take a cue from how they do their separations. I’m not doing that and I haven’t (I don’t need anything Fulcro can’t provide), but I’ve built projects in similar ways in the past.

Thank you for the elaborate answer! A desktop app together with a React Native app was what I am planning to go for. All clients would be controlled so that wouldnt be a problem. So I guess the Fulcro/GraphQL route sounds promising for that usecase

You don’t even need to use Graphql. You should be able to do it all in Fulcro with eql then. React Native and Fulcro has been a very pleasant experience so far.

Oh really? I thought the React-Native story with ClojureScript was still a bit wonky last I checked. Good to hear that its a pleasant journey! I may look into that

I haven't hit the issues yet. I was worried about performance and app size but right now the app checks in at 11mb and it's buttery smooth. Probably smaller than other RN apps because I'm less inclined to use libraries (though it's extremely easy to do)

And the idea of not writing it in clojure at this point is... Unpleasant

granted that’s a pretty minor issue 😄

idents go both ways: used to ask what the component’s ident is on things like refresh, and used for normalization. In union case, the parent is used for normalization, and the child is the active thing on screen…thus the need

There’s certainly a macro waiting to be written there, but the EQL nesting of it needs two components, even if you don’t much care about the parent

but perhaps I’m not understanding your question

Sorry, I’m probably not being very clear, it’s such a simple thing probably not worth your time… 🙂

(defsc UnionComponent
  [this props]

  {:query
   (fn [_]
     {:type1/id [....]})

   :ident
   (fn []
      (cond
        (contains? props :type1/id)
        [:type1/id (:type1/id props)])}
...)

(comp/get-query union-ns/UnionComponent)

(defn get-ident
  [props]
  (cond
    (contains? props :type1/id)
    [:type1/id (:type1/id props)]))

(defsc UnionComponent
  [this props]

  {:query
   (fn [_]
     {:type1/id [....]})

   :ident
   (fn []
      (get-ident props))}
...)

((first (get-ident props)) (comp/get-query union-ns/UnionComponent))

Which is ultimately fed into:

(dnz/db->tree query concrete-props state)

Regardless, I think I’ve obviated the need to manually get the local state tree… I’m refactoring part of an existing application.

Didn't we arrive to that you need to set the right CSS prop on the component for it to respect \n? So just somehow eg give it a particular CSS class (I think there is ao/<something> for it now) to it and define the CSS class with the prop?

white-space - https://stackoverflow.com/a/30593806/204205

I second that; I’ve done this stuff inside a Clojurescript text editor built on textarea; even then it’s just CSS

Thanks for the reminder. Done.