The Changelog: Software Development, Open Source - An API store for LLMs, DeviceScript, Nyxt: the hacker's browser, expectations debt & there's still no silver bullet (News)

Episode Date: May 30, 2023

The Gorilla team is building an API store for LLMs, DeviceScript is Microsoft's new TypeScript programming environment for microcontrollers, Nyxt is a hackable browser written in Lisp, Morgan Housel writes about expectations debt & I issue a gentle reminder to my fellow software engineers: there's still no silver bullet.

What up, nerds? I'm Jared, and this is Changelog News for the week of Monday, May 29th, 2023. Only it's coming at you on Tuesday, because Memorial Day. Can you believe that Changelog News only has one rating in Apple Podcasts? At least it's a five-star rating, but still. If you're listening in Apple Podcasts right now, hook a brother up with a five-star review. Unless you're listening to this in your shower, that might be dangerous. And don't forget to clean behind your ears.

Okay, let's get into the news. The team behind Gorilla is building an open-source, one-stop shop for all APIs that LLMs can interact with. The problem is, even state-of-the-art models like GPT-4 are prone to calling the wrong API, especially for more obscure languages. For example, Google BARD is batting zero for me on a dozen or so Elixir-related API requests. Enter Gorilla. They say, quote, given a natural language query, Gorilla comes up with the semantically and syntactically correct API to invoke. With Gorilla, we are the first to demonstrate how to use LLMs to invoke 1600 plus and growing API calls accurately while reducing hallucination, end quote. This is a big step in a good direction. The team behind Gorilla is laser focused on fixing this

problem and they've even developed and released a dataset called API Bench for evaluating the model's results. DeviceScript is Microsoft's new TypeScript programming environment for microcontrollers. It's designed for low-power, low-flash, low-memory embedded projects, and has all of the familiar syntax and tooling of TypeScript, including the NPM ecosystem for distributing packages. This project has a lot of devs excited. Jonathan Berry says, quote, Always glad to see these attempts at bringing web technologies to embedded systems and see what sticks. Even when they don't, they inspire innovation.

Zach Silviera says, quote, And Andrea Ghiomarchi says, quote, this is so much better than MicroPython. And Andrea Giamarchi says, quote, this is the first Esperino competitor, and I think it's going to be huge. Have you ever heard of Nxt? N-Y-X-T? They call it the hacker's browser, and it's written in Lisp, so you know it's good. It's made with bits of real panther, so you know it's good. It's made with bits of real panther, so you know it's good. Out of the box, Nixxed ships with tens of features that allow you to quickly analyze, navigate, and extract information from the internet. Plus, Nixxed is fully hackable.

That means all of its source code can be introspected, modified, and tweaked to your exact specification. Check the project's homepage for some pretty compelling features. It has a command prompt that provides fuzzy search, link hinting, a REPL, clipboard history, and a whole bunch more. It's now time for some sponsored news. Stuck while coding? Need some copy? Imagine this, you hit command space

and you're already asking ChatGPT anything. Enter Raycast, a blazing fast, totally extendable launcher for macOS. Think Spotlight, Quicksilver, or LaunchBar, only better. Today, the Raycast team is launching a pro offering right here on Changelog News. This thing has so many goodies, it's hard to list them all. AI, Cloud Sync, Custom Themes, Unlimited Clipboard History, Pro APIs, and more on the way. Use Raycast for absolutely free and upgrade when you're ready to go pro.

The link is in your show notes. Thanks to our friends at Raycast for sponsoring ChangeLog News. Morgan Housel writes about living in Seattle when Amazon was at the top of the world not so long ago. Quote, expectations were so high in 2021 that investors and employees had to achieve extraordinary things just to break even. When results were merely good, they felt terrible.

Expectations are like a debt that must be repaid before you can get any joy out of what you're doing. The hard thing is that every company and every employee wants to have what Amazon had in 2021. Winning, wealth, prestige, reputation. But look at what it led to now. After the expectations debt was repaid, was it worth it? Hard to say. End quote. We talk a lot about technical debt, but this is my first introduction to the idea of expectations debt. Setting expectations is so important in many aspects of life, but sometimes you don't get to control those expectations. Example, you're a movie director and someone hears from their friends that they absolutely must see your movie because it's life-changing and the best thing ever.

So they go to see the movie with high expectations, and it doesn't live up to the hype. That's outside of your control as a director. Market dynamics are also outside of our control. So how do we manage our own expectations as we ride the waves up and down. Something worth thinking about. What if, uh, C-A-T really spelled dog? Wow. Yeah, that's heavy, Ogre. I've found myself referencing Fred Brooks's No Silver Bullet paper a couple times recently. Here's one of them from our JS Party episode with Dan Abramov and Joe Savona from the React team. I think when we talk about SPAs in the large, it goes back to Fred Brooks' No Silver Bullet, you know? And we tend as a ecosystem to point out the next silver bullet and then use it for everything.

Or the old saying, like like when all you have is a large hammer, everything looks like a nail. Nick, you've known me for a long time. I've been talking about the right tool for the job, like the entire time as we swing the pendulum back and forth. And that's a difficult conversation to have because in a sense, it's kind of a cop out, like, well, it's easy thing to say right tool for the job, but it's a hard thing to actually select the right tool for the job. What's easy is to find a great tool and then use it for all of your jobs. And then you realize that actually it's great at this specific thing and maybe it's a hard thing to actually select the right tool for the job. What's easy is to find a great tool and then use it for all of your jobs. And then you realize that actually it's great at this specific thing and maybe it's okay at that other thing,

but it's not perfect. Now all of a sudden we're just, you know, hitting ourselves on the, on the thumb with this large hammer trying to use it wrong. And then we get mad at the tool maker eventually like, Hey man, this tool doesn't do all the things that I want it to do.

And it's like, yeah, it doesn't. That conversation was in the context of React, but the principle applies to every over-adopted technology. So I wrote a post as a gentle reminder to my fellow software engineers. If a new tool promises to solve all your problems, it won't. Look at this. Yes.

Combination hookah and coffee maker. Also makes julienne fries. Will not break. Will not. It broke. If you think you found the one true tech, you haven't. And if you have a shiny hammer that looks great for hitting screws, it isn't. That is the news for now.

On Wednesday's interview show, we are closing out Maintainer Month with some more awesome conversations from Open Source Summit. And on Friday's talk show, we are kaisening once again with one of our oldest internet friends, Gerhard Lezu. Have a great week.

Share Changelog with your friends who might dig it. And I'll talk to you again real soon.