Dustin Getz

I am imagining (hf/valueType ctx) (etc) that infer the fiddle-ident from the ctx

https://forum.datomic.com/t/lookup-ref-upsert/826

It should just fail fiddle/ident validation because this is always an accident. when the day comes that someone actually wants a ! in their keyword we can talk to them...

Instead of escaping can we fallback to percent encodings?

I see, I think & would work

Not actually sure what's wrong with it yet ``` (let [{:keys [:hypercrud.browser/fiddle]} ctx] [:div props [hyperfiddle.ui/markdown (:fiddle/markdown @fiddle) ctx] (->> val (sort-by first) (map-indexed (fn [i [t username]] [(.toLocaleDateString t)...

``` [:in $users :find ?t ?name :where [$users ?user :user/user-id] [$users ?user :user/name ?name] [(user/entity-creation-tx $users ?user) ?t]] ```

Ok I confirmed - localStorage.clear() fixes, and then restoring this query and render reproduces it in subdomain "d" on my machine, i will have to set it up in prod...

It is the partition-by first which is broken - it causes reagent to try to interpret a string as hiccup, with no error boundary i guess

I think we need this in table-picker, I don't immediately see where it goes so leaving for you. Reverting commit for now to uncrash it ``` ::hf/ident-key (if is-ref ;...