anybody have a favorite way to chunk code? ignoring the more boilerplatey stuff i've been putting my components in a ui folder and i have a models folder where i put pathom resolvers and mutations in .clj files and fulcro mutations and uisms in .cljs files. it feels OK so far but my project's still pretty small

Our little fulcro study group, is meeting again in a few mins =)…

Is there an example of Fulcro 3 dynamic routing + Bidi (for html5 history)?

Hi, my app ran fine until yesterday. Today I updated deps from shadow-cljs 2.10.12 to 2.10.14. rollback solved it once. Then I tried to make the error reproducable. Now I can not get it to work again. I cleaned .m2, ressources/public/js, target, classes. My builds on the template. calling routing to start in init starts the errors. If I comment this out then mounting the app calls the errors. The state machine of the dynamic router won't be loaded. All the errors are similar.

ERROR [com.fulcrologic.guardrails.core:54] -  Error: Unable to resolve spec: :com.fulcrologic.fulcro.ui-state-machines/fulcro-app
    at Object.cljs$spec$alpha$reg_resolve_BANG_ [as reg_resolve_BANG_] (alpha.cljs:71)
    at Object.eval [as cljs$spec$alpha$Specize$specize_STAR_$arity$2] (alpha.cljs:137)
    at Function.eval [as cljs$core$IFn$_invoke$arity$2] (alpha.cljs:132)
    at Object.cljs$spec$alpha$specize_STAR_ [as specize_STAR_] (alpha.cljs:131)
    at Function.eval [as cljs$core$IFn$_invoke$arity$2] (alpha.cljs:161)
    at eval (core.cljs:4741)
    at eval (core.cljs:4742)
    at Object.eval [as sval] (core.cljs:3439)
    at Object.eval [as cljs$core$ISeqable$_seq$arity$1] (core.cljs:3496)
    at Object.cljs$core$seq [as seq] (core.cljs:1234) BUG: Internal error in expound or clojure spec.

When I eval the errors read a bit differently than in the log. Evaling app/mount returns #object[Error Error: Unable to resolve spec: :com.fulcrologic.fulcro.ui-state-machines/env].

Has anyoune seen this before?

@magra So, if you disable spec generation in guardrails this can happen…it is a combo of options that should not be supported, actually.

there is an emit-spec? option, which should be set to true if GR is enabled

you can also change the GR option throw? to be false so it just emits warnings instead of throwing (which is what I do in dev).

I only use throw? true in tests (because there is never a good reason in my experience to throw in clj/cljs due to a spec failure that already can print a stack trace)

More in depth: >def is like s/def, but can optionally NOT emit a spec. This is to make cljs builds smaller…however, if you tell it not to emit specs and also turn on spec checking, then you’ll get that error. Not sure why the upgrade gets you, but I suspect your deps tree changed….I thing G.R. had a bug where it was ignoring the emit option or something that got fixed, and you probably just got a newer version.

this is unrelated to shadow itself

@magneh i think you have to switch this to true

https://github.com/fulcrologic/fulcro-template/blob/master/guardrails.edn#L3

Wow! fixing it in guardrails.edn was not enough. I had to change shadow-cljs.edn from the empty map to {:throw? false :emit-spec? true}.

now that is strange. That map should override the edn file.

Thank you Toni!!!!

If this is not specific to my setup, but a new dependency then the template might trap up newcomers from now on.

I’ll update the template…it’s long overdue

Thank you again. I searched for a few hours and would have wasted another day since I had ruled that out and was barking up the wrong tree.

I tried old Versions from a week ago so I had ruled out a misconfiguration.

So I had an invalid guardrails.edn with a good override from the start and something stopped this override from working.

I’ve updated the Fulcro template: https://github.com/fulcrologic/fulcro-template to latest dependencies and tested everything (I think). I had no problem getting GR from config file to work properly.

I’ve added a comment to the README about how the RAD demo might be a good starting point these days instead of the low-level template.

RAD just gives you more…but all of it is optional. But I can’t imagine why you wouldn’t want at least some of it.

I must have missed some stale state then after changing the .edn file.

https://clojurescriptpodcast.com/ Part 2 of Fulcro discussion on cljs podcast is out today it seems

Am I crazy, or is there something weird with singletons (constant :ident) and workspaces?

Any time I add a constant ident to a component that is at the root of a defcard the ui disappears (both with and without :wrap-root?)

initial state?

if it has an ident, you have to use wrap root true

and you need to have initial state for it

(defsc cool [this {:keys [] :as props}]
  {:query []
   :ident (fn [] [:component/id :cool])
   :initial-state {}} ;; tried both with and without initial state being empty
  (dom/div "HI"))

(ws/defcard CoolCard
  (ct.fulcro/fulcro-card
    {::ct.fulcro/root       cool
     ;::ct.fulcro/initial-state {}
     ::ct.fulcro/wrap-root? false ;; tried both with and without :wrap-root?
     ::ct.fulcro/app        {:optimized-render! kr2/render!}
     }))

All 4 scenarios are blank for me

you didn’t query for anything

why give it an ident if there is nothing to get, and if there is nothing to get, the wrapper might short-circuit….I didn’t write workspaces, so I don’t know internals of what it does…

Ok. I might dig deeper here to see what's up. If I find that it's a bug in workspaces I'll make a pr against ws3

(adding a query doesn't seem to help fwiw)