Clojurians Log v2

just foo is enough instead of ((fn [_]) foo)

(defmacro eager [v x & body]
  `(let [~v ~x]
     ~v
     ~@body))

I'm intrigued that this is possible

(defmacro eager [v x & body]
  (if (symbol? v)
    ;; simple, non-destructuring version
    `(let [~v ~x]
       ~v
       ~@body)

    ;; destructuring version
    `(let [value# ~x
           ~v value#]
       value#
       ~@body)))

(defmacro do1 [expr & rest]
  "Sequence like `do`, but return the value of the first expression."
  `(let [value# ~expr]
     value#
     ~@rest
     value#))

(e/defn Summary-Detail [Summary Detail]
  (dom/div (dom/props {:class "flex flex-col"})
    (eager expanded
      (dom/div (dom/props {:class "flex flex-row"})
        (do1
          (dom/div
            (dom/props {:class "w-6 pl-1"})
            (Expanded-Icon))
          (dom/div
            (Summary))))
      (when expanded
        (dom/div (dom/props {:class "flex flex-row"})
          (dom/div (dom/props {:class "w-6"}))
          (dom/div
            (Detail)))))))

right, it's just functions 😉

let bindings are lazy, they won't run unless used. (let [_ (prn 'foo)]) will never print foo. Put the debug prints in a do (e.g. in the let body)

ah, but won't they run if a value is passed that changed?

if the binding (in this case _) isn't used somewhere, no

ahhh, so it's about the left-side binding

I would have expected it to run if the right-side touches a symbol that changed

yes, let basically compiles to a missionary signal. The signal won't be sampled unless used

thx this fixes a big misunderstanding I had

(deleted snippet to not pollute channel - same component is incluced as snippet in next message on new thread)

@dustingetz in meantime until I can make smaller example, if you wrap https://github.com/theronic/eacl-electric-starter-app/blob/main/src/electric_starter_app/main.cljc#L393`e/for`https://github.com/theronic/eacl-electric-starter-app/blob/main/src/electric_starter_app/main.cljc#L393 in e/server, you should be able to replicate the crash. it seems to render the first time but fails on 3rd re-render, which implies a potential diffing issue 1. wrap the e/for in e/server 2. click Next Page twice (regardless of page size) 3. reactor crashes.

Here is https://github.com/theronic/electric-efor-bug-repro/blob/main/src/electric_starter_app/main.cljc of e/for crashing Electric inside an e/server (but not e/client) when looping over 1k plain maps that are sorted and paginated. Was forked from v3 starter app (no Datomic, no EACL, no Mount). Also fails if you click Next Page button followed by Prev Page button.

I’m compelled to say this is an e/for bug in Electric. Same bug bit me in another project today: Reactor crashes with no recourse. Have to restart REPL to recover and no error logs. Solved by taking e/for out of e/server and running on e/client. Will reproduce minimal example and share.

1st & 2nd pages look like this, 3rd page is blank (client crash)

nothing jumps out as immediately wrong on a quick skim. I'd try minimizing the breaking example by commenting out parts of it

we will fix it but we need a way to reproduce it, this example is not self contained

if you give us a qry-accounts with mock data (inine the actual data as edn) and it still crashes, that might be enough

try moving the log to after the e/for, because electric is concurrent and is probably running those expressions in reverse, i.e., you see the log for the record before the record that crashed (if it is the record which is bad)

i know you said the order seems not to matter, but still the diff-by is a likely culprit

Hello 😄 Reagent expects you to install React: https://github.com/reagent-project/reagent?tab=readme-ov-file#usage re-graph describes itself as "GraphQL client for re-frame applications", you might want to try using a GraphQL client which isn't coupled to another ClojureScript framework? In any case this is probably a better question for #beginners or #clojurescript

ok - thanks - I considered starting there, but thought it might be specific to electric, as re-graph has worked fine for me in regular (not clojurescript) projects in the past

In these projects you presumably had react installed already?

I dont think so - and I think I may have figured out my issue - I found some :require entries (in dev.cljc) that are prefixed with "#?(:clj" - copying that (macro invocation?) in my electric component with re-graph gets me past the build error!

I think you're treating the symptom not the problem now :^)

Wrapping in #?(:clj ...) makes it so that the wrapped form is only read (and evaluated) on JVM Clojure.

could be! (and thanks for the help btw). re-graph says its "Clojure/Clojurescript" so I guess I thought it was a package that could be build (perhaps with different feature-sets?) in either context

but again, I am way out of my depth here

Oh, right I missed that

So if you do #?(:clj (:require [re-graph.core])) you should be able to use it on the JVM side no problem

https://clojure.org/guides/reader_conditionals

No problem 😄

ok, awesome, I've yet to make the call, but its nice to know that I could reasonably expect it to work 🙂 (on the jvm side)

may I ask another question? would I be right in thinking that I cannot call an electric function within e/Offload ?

(that is to say, I get a macroexpand error when I do this)

Correct

found issue: caused by an error in a function running in (e/Offload #(...))

Thank you for the report 👍