Building #ElectricClojure and https://xitter.cf/t.co/JkYr6J4dpu. I believe in excellence, and I believe that many others do too. Baháʼí. https://xitter.cf/t.co/aBmRZmuoCs
🌎
Joined May 2009
- Tweets3.1K
- Following3.2K
- Followers4.6K
- Likes16.2K
Dustin Getz retweeted
People still don't understand how the mental model for building AI agents is closer to a frontend than a backend service
The call graph is dynamic, depending on what the LLM or user saw / decides. They need a BFF that shields them from service topology and product joins.
Dustin Getz retweeted
Thank you Guillem
I believe development that focuses on user experience and rich features is what enables this fast pace.
Technically the dev speed is enabled by #Electric from the Hyperfiddle team (led by @dustingetz) and the associated Electric-specific LLM skills.
Yesterday we had the first Incremental Learners community meetup. @adham_clj shared the most important updates of freememo.net.
The development speed is incredibly fast-paced for a solo side project. Just the July updates feel like more than 4 years of a SuperMemo big release!
Dustin Getz retweeted
Endive 1.0 is out!🚀🚀🚀
Pure-Java WebAssembly runtime, zero native dependencies, now a @bytecodeallies project.
Now with native JVM GC integration for WasmGC references and tail call optimizations!
Chcek it out: endive.run/blog/endive-1.0
Dustin Getz retweeted
Very impressed with Clojure Electric v3.
Tried v2 a long time ago and found it very hard to wrap my head around the bad error messages.
Now, especially with help from AI, it's very easy to write apps with a fraction of LOC than a react or clj program.
Dustin Getz retweeted
github.com/hyperfiddle/elect…
久しぶりに触って前より理解できる。solidjsやsvelteのようにシグナルの値とそうでない値が混在するのではなく、全ての値がシグナルならリアクティブを意識する必要もなくなるって感じか。
同じ関数内にクライアントサイドのコードとサーバーサイドのコードを混在させられるのはやっぱすごい
Dustin Getz retweeted
there really isn't anything out there like Missionary (by @36Rleonoel) if you want dynamic reactive computation at runtime with async & I/O built-in
Dustin Getz retweeted
broadly we frame UI as a fine grained resource management problem, which is a high frequency concurrency problem - during a drag drop or virtual scroll we are updating, allocating and disposing resources at the animation rate (120 hz) and many of the resources are remote.
Re transactional semantics - we see this in forms and optimistic updates (todomvc for example - the create new operator should apply optimistically and rollback if server rejects).
these two worlds are two different concurrency models - transactions are discrete event streams ( happen at point in time), and resource management is (surprisingly) continuous state signals (values that are always defined, the resource effect lifecycle i.e. mount/unmount comes for free from the state transition change stream). The backpressure semantics are also different in these two worlds - discrete events are push/eager and continuous states are pull/lazy.
Resolving this impedance mismatch (finding a common parent structure and combinators defined for both continuous and discrete computations, so that our UI can have a uniform composition model that spans the end to end computation) is the central problem of UI in our view and this is what Electric is trying to achieve.
Dustin Getz retweeted
one of the bigger problems in enterprise AI right now is that because leadership has already told the world that AI is working, the people trying to make it real aren't allowed to admit that they're struggling. and can't go through the trial and error required to learn
Dustin Getz retweeted
Type Checking is a Symptom, Not a Solution open.substack.com/pub/progra…
Dustin Getz retweeted
Dear founders I have a simple protocol for how to pitch a VC:
**YOU MUST DO THIS BEFORE THE PITCH**
> go 2 Home Depot
> back right corner
> buy a good size 2x4
> drive to VC office
>> BEFORE WALKING IN, hit yourself in the head w/2x4 at least six times
> now you are ready
Dustin Getz retweeted
Hiring a senior Clojure engineer at CloudAfrica. Full-time, remote.
Networking (ISP) experience preferred. Electric Clojure experience is a plus.
Our stack is Clojure + Datomic + Electric Clojure + Proxmox + Qemu.
Dustin Getz retweeted
watching infrastructure companies being forced to raise into AI hype to stay alive is incredible
Dustin Getz retweeted
I've got this silly idea that I can't shake that LLMs will be better product managers than 99% of professional managers simply by prompting them to use product management frameworks. Everything one needs to know about management can fit in a single book and yet ~all startups fail
Dustin Getz retweeted
The recent European Lisp Symposium had a talk on whether Lisp is still relevant in the age of AI. Many, myself included, found it somewhat lacking. @danshipper however inadvertently nails it: "Your job isn’t to type code anymore, but to design the systems that design the systems." Isn't this precisely the lisp credo of solve a problem by writing a language in which the problem is easy readily expressed and solved and also Norvig's Systems Paradigm.
we have completely changed our engineering philosophy @every because of Claude Code
we called it Compounding Engineering:
Each feature should make subsequent features easier to build, not harder.
@kieranklaassen just wrote THE definitive guide to each step of compounding engineering today on @every:
Dustin Getz retweeted
I don't think every UI will be generated on the fly, but apps will want to be more and more customized per user
the problem is our current stack is terrible at this
serving 1 site to 100k users is a very different problem to serving 100k slightly modified sites
Dustin Getz retweeted
BY FAR the biggest reason I'm souring on React is the discrepancy of what's happening in the DOM and how I express my code
I just want to do something on mount, just want to create an instance once and never again, or 100 other things
React has built it's own language so you have to model these things in an unfamiliar way. `useMemo` doesn't actually fully memo, effects can be called multiple many times despite what you want, and many other surprising behaviors. Their language has unlocked interesting things, but it's nothing I'm using but I still have to pay the cost of confusing code
Sure, I could use a framework on top of it to help (I do, old remix) but I still write components. Implementing anything complex always ends up with strange workarounds
Dustin Getz retweeted
Zero-allocation, type-safe state management is possible on the JVM with Scala 3.
Kyo leverages this in `Safepoint`, an internal mechanism for stack safety, thread isolation, and debugging. Thread depth, interceptor presence, and ThreadId are all tracked in a single `Long` via bit packing.
Extension methods make access + updates trivial. No classes necessary.
ALT opaque type State = Long object State: // Bit allocation: // Bits 0-15 (16 bits): depth (0-65535) // Bit 16 (1 bit): hasInterceptor flag // Bits 17-63 (47 bits): threadId private inline def DepthMask: Long = 0xffffL private inline def InterceptorMask: Long = 1L << 16 private inline def ThreadIdMask: Long = 0xffffffffffff0000L require(maxStackDepth <= 65536) inline def init(): State = (Thread.currentThread().getId() << 17) & ThreadIdMask extension (state: State) def depth: Int = (state & DepthMask).toInt def threadId: Long = (state & ThreadIdMask) >>> 17 def hasInterceptor: Boolean = (state & InterceptorMask) != 0 def incrementDepth: State = state + 1 def decrementDepth: State = state - 1 def withInterceptor(hasInterceptor: Boolean): State = if hasInterceptor then state | InterceptorMask else state & ~InterceptorMask
Dustin Getz retweeted
Excellent thread on Mastodon taking a deep dive into effect systems, continuations, and how we can improve on the state of the art in Scala. fosstodon.org/@raulraja/1095…
