auth architecture of your app requires it, not because of a 3rd party mechanism - WebAuthn.

Nice!

But coming back to the DNS part, short of service-worker caching the (auth part of your) app - or unforgivable hacks like installing a local DNS-resolver - there is a hard requirement for internet connectivity on (first or empty-cache) auth.

When ideally that would only be needed when the …

I thought WebAuthn was limited to online apps (due to its reliance on DNS), how could you use this for local-first apps when there is no DNS name…

Is there a trick I’m missing?

But… what about those poor datalakes?

Without someone managing the datafountains, how do you keep your data from going stale?

Can write a DSL for it later if desired.

Why not pass around a vector/list type which implements function calling, which works like a thunk/partial when called.

Constructing & more partial application via a fn (or macro), so starting from a bare fn or an already constructed vector-type works identically.

Best of both worlds?

But it also is an old language, with evolved syntax… nice for backwards compatibility, but some stuff is very obviously bolted-on-later 😅

Actually™️, Python has the second best function API-spec (argument definition) of dynamically typed languages I have encountered. Only Clojure’s is better IMO.

Keyword arguments, default-values, wildcard args & kwargs, and (admittedly strange) syntax prevent overlapping args & kwargs if so desired.

Ah, so that's why I can't find the Linux download anymore.

I assume it will stop working altogether for Linux until you revisit platform support for it?

Getting Real by 37signals basecamp.com/gettingreal/... has a nice section just for ideas. It sees them as a multiplier for good execution, but not of much value in and of itself.

… or go the Lisp route and _almost_ keep both.

With for the JSX example (going with ClojureScript here):

- Hiccup for html-ish syntax
- Reagent for React to “something nicer” (especially true for pre-hooks React)

The consession to syntax is CLJS-native datastructures as syntax-primitives.

This one by Peter Alvaro is an oldie, but very good - one of the parts of research that went into hydro - imagine you favorite teacher speedrunning distributed systems and casually presenting a beautiful solution to a very hard problem. youtu.be/R2Aa4PivG0g?...

Reminds me of johnsalvatier.org/blog/2017/re... …

I stumbled upon this while reading Self papers - its old, probably a very bad idea (RMI & Distributed Objects based distributed system), but interesting nonetheless: www.ag-nbi.de/research/dse...