I bit the bullet and went towards re-doing the effects system implementation with the native async/await, to see if I can get a baseline implementation. I was successful, but not without roadblocks along the way. This version is much simpler and handicapped than what I wanted originally, but I
These weeks are passing by faster and faster. I too would like to work faster, but I didn't think vibe coding was going to be it. I work with AI everyday, and I mostly have it do basic debugging or explaining Python minutiae to me. I know what
❓This is a conversation with GPT o3-mini💬What is a Kleisli Category? Explain it to me as someone that knows functional programming well. A Kleisli category provides a way to encapsulate and compose effectful computations, which is especially appealing if you're familiar with functional programming and monads. Core
The work on the scheduler is to support async effect handlers. It won't do at all, if a network request always blocks the evaluation of the computational graph. So I embarked on it about two weeks ago. Right now, the API looks like this (subject to change). There&
This was a tough past week. I was working on supporting async handlers, even if the reactive part was synchronous. I didn't want the graph to have to wait on network requests. This turned out to be tough. Also I got food poisoning this past week. The design
There are some programmer that love types and there are others who eschew them. When reasonable people fall on both sides, it's good reason to look for a latent variable that better explains the underlying motivation or cause. Latent variables, by definition, are hard to measure or observe
I made good progress on the reactive engine. * Simplified the requirements for define a handler. * Started work on the scheduler * No longer need to handler to ability type checking method, by converting from protocols into abstract base classes. This was one example of where AI lead me astray. I also
A couple of months ago, I had to reload everything I knew about reactivity back into my head. I discovered that I missed a large corner of the development around unstructured reactivity, by way of Signals. From arguments about signals vs. functional components on Twitter and in post comments, it
❓This is a Deep Research report with GPT o3-mini💭<goal> If I wanted to build and design an algebraic effect system in a language that didn't support delimited continuations (like python), but I wanted multi-resumption, I wouldn't be able to use generators. So how
Underlying great creations that you love—be it music, art, or technology—its form (what it looks like) is driven by an underpinning internal logic (how it works). I noticed this pattern while watching a talk on cellular automaton and realized it's "form follows function" paraphrased
I spent a good part of this week writing and rewriting a blog post, as part of my commitment to publish more this year. I wrote about 10k words, only to keep about 4k of them. Must be the sunk cost fallacy, because it feels bad to delete entire pages
With the release of OpenAI's Deep Research and its testimonies on Twitter, I finally slapped down $200/month for Pro subscription. Deep research is what I do all the time by hand anyway, I figured I'd get a lot of mileage out of it. The second
ai-chat
❓This was a report from GPT3o-mini with Deep Research💭What are the different ways to do handler look up in programming languages that support algebraic effects? I know sometimes, they're also called "abilities" in some languages. A starting point of languages that I know support algebraic
lab notes
I'd finished up looking into reactivity (with the exception of data type enforced reactivity, like differential dataflow), and I pulled out code with pedagogical sample code implementing signals and autotracking. * iamwilhelm/pedago_signals: A pedagogical implementation of hybrid push-pull signals in Python * iamwilhelm/pedago_autotracking: A pedagogical implementation
ai-chat
❓This is post is the result of a Socratic conversation I had with GPT-4o about what "syntax represented by the free monad for a functor that provides a signature." meant. After the question and answering, I asked it to write an outline of all that I learned and
I almost never remember what I did in a year. Having written reviews for 2021 and 2023, I've found it helpful to both have a record to look back on as well as a point to review and reflect. I think this should be a habit every quarter,
ai-chat
❓Oleg Kiselyov has a post and corresponding talk, "Having an Effect", about how to represent side effects in a language. While I watched the talk, I conversed with GPT4o about any questions I had. At the end, I had it write a blog post summarizing our conversation, with
lab notes
I'm a little indisposed this week, but I did find some time here and there to do research into effects. I spent the time reading and watching a talk about effects by Oleg Kiselyov. The short version is that side-effects can be formulated as a communication protocol between
lab notes
I'm back, after a little hiatus from writing the notes. Last time, I discovered that I should have been looking more deeply into other reactive models before I embarked on a functional component (React-like) model to model data pipelines. The impedance mismatch should have been indicative that something
lab notes
Last week, I started looking at Signals. When I stepped back and looked at the entire front-end ecosystem, it seemed like many of them outside of React (Solid, Vue, Svelte, Preact), converged on some variation of Signals. There is even a Javascript proposal to include Signals in the language. Given
lab notes
One of my main goals is to make non-CRUD backend pipelines more visible. I didn't think nodes and wires programming was the way to go, but unconsciously, I envisioned just showing the results of a cell next to the cell, like regular computational notebooks. But my underlying assumption
Essays
🔰If CRDTs (Conflict-free Replicated Data Types) are new to you, check out this previous post. There's a simple way to understand CRDTs: It leverages algebra to unmix the inevitable mixing of data when syncing over an unreliable network. Why are networks unreliable? For one, it'd be
lab notes
I've been a bit anxious about the pace of progress these last three weeks. I wanted to strike a balance between making practical progress and having a sound basis. This last week, I struggled with breaking from structured programming. * What do algebraic effects buy you? * What is the
startups
Despite the disdain for "GPT-wrappers," it's currently where the value is accruing in the supply chain of value to the end-user with LLM-driven apps. Unless something changes with the foundational model companies approach the market, the GPT-wrapper application layer is where the value accrues, hence where