Clojurians Log v2

consider putting little $50-$500 bounties on stuff like this and people will do it mostly for street cred. pay in crypto, make a big hoorah about it. ouks will poast about it and amplify Electric. make it feel like a leetcode challenge. “who can build this the fastest in Electric? time yourself. record a livecoding sesh if you want.”

(tax-deductible marketing expense)

Got some experience with Slack messages. It's not a particularly pleasant place. Partly because you have to constantly be translating between users and their user ID. Clojure does make it easier.

And I have not even mentioned the trouble around permissions for the bot. You have to approve every single thing that the bot is allowed to do. The owner of the namespace has to allow the bot and the permissions on that bot, as well as presumably added to some channels.

Changing how it is viewed by humans, such as impersonating a human, requires some serious permissions. I wonder whether it's even possible to do so in the Clojurians' Slack because you're not an admin of the Clojurians' Slack.

Prior art https://github.com/emacs-slack/emacs-slack

Most likely electric lost a type hint during compilation. A way to work around this is to force clj compilation, i.e. extract to a cc/defn or use an anonymous fn

Good to know, thanks! Just wanted to make sure I wasn't missing something deep about the way that Electric treats non-reactive collections.

generally speaking, in this expr, ns (range 1 (inc n)) since n is variable, you probably want ns to be incrementally maintained, which means you should diff-by asap, certainly before mapping squares (map #(* % %) ns) because that will recompute everything when ns changes incrementally. In other words, in Electric code bodies there is very little use for for & map (and clojure sequences broadly), you almost certainly want to be using e/for

Thanks, that makes sense. I actually came across the behaviour when I was working through the fizz buzz tutorial and experimenting with what happens when you replace the reactive/diffing collection with a regular collection just to understand the difference. I really like the diff-by syntax that turns a regular collection into a diffing one. Only thing I find is that sometimes I have a hard time keeping a track of what’s a normal collection and what’s diffing, but your advice to diff-by asap makes sense.

yeah, the answer is "every most collections should be immediately diffed as soon as electric touches it" which means "plural" names like xs are diffed. I sometimes write xs! to mean non-diffed in advanced cases

an immediate exception to "diff everything immediately" is tables, where you want to retain the original clojure collection (may be large) in memory and the virtual scroll will encapsulate the rendering/network strategy

because here, what you really want is to stream just the collection's viewport window to the browser, so you're diffing just the window, not the entire collection

But if I try the same sort of thing using for then I get a compile error

(e/defn Sequences2 []
  (e/client
    (let [n (parse-long (Input* 10 :type "number" :min "0" :max "100"))
          ns (range 1 (inc n))
          squares-using-for (for [x ns] (* x x))]
      (dom/div (dom/text "n: " n))
      (dom/div (dom/text "ns: " (pr-str ns)))
      (dom/div (dom/text "squares-using-for: " (pr-str squares-using-for))))))

Actually, here's an even more minimal example:

(e/defn Sequences3
  []
  (e/client
    (dom/div (dom/text (pr-str (for [i (range 10)] (* i i)))))))

https://github.com/clojure/clojure/blob/master/src/clj/clojure/core.clj#L4748

Thanks for the link, but I'm not sure I follow. Are you saying that the expansion of the clojure core for macro somehow interferes with the way that electric works?

I dunno, just the for macro expansion is very complicated, way more than calling map

The line I linked to is almost identical in the clojurescript for definition, and I bet that is actually the issue, maybe related to the ^not-native annotation in the cljs

Isn't it in the same repo? https://gitlab.com/hyperfiddle/electric-fiddle/-/blob/main/src/electric_tutorial/typeahead.cljc

Thanks! Not sure how I missed that

how did that go? I forgot we use that so early in the tutorial

we definitely don't intend for you to understand the typeahead impl that early