Chad Adams

First time using Claude Code and what a pile of junk.

The vlang setup-v github action is now ~30x faster on subsequent runs when the version is specified. github.com/vlang/setup-v/…

@BenjDicken Just out of curiousity but how many connections can Postgres handle at once before this becomes a problem?

Postgres is bad at handling lots of connections. It's one of the biggest pain-point I've seen from customers over the past year. PgBouncer is the way to solve this. It's powerful, but also easy to misuse. So, I wrote about it.

The vlang lsp now supports doc comments on hover

@brk0v Zero-cost abstractions means you have a slow compiler.

🦀 Rust zero-cost abstractions aren’t always zero-cost. Turbopuffer chased a nasty perf cliff and found Rust iterators were silently blocking SIMD in a hot merge loop. Result: query dropped from 220ms to 47ms. turbopuffer.com/blog/zero-cost #rust #rustlang

@GregorySchier Yeah that must of been a long time ago 😆 would highly recommend trying it again. I also use Rust, Go, TypeScript. Haven’t had any issues. I’ve tried Zed can’t get into it. For some reason the UI feels weird to me. Maybe I’m just so used to VSCode 😆

@cadamsdev Haven't used it in many years. I remember extensions crippling performance, especially because I wrote Rust, Go, and Typescript on the daily (lots of extensions)

I switched to Zed from IntelliJ a couple months ago due to how buggy it got. Now Zed is loaded with bugs so... back to IntelliJ!

@notyuldshah @OrenMe @GitHubCopilot @burkeholland @code I prefer Sonnet 4.6 because it's faster and cheaper than Opus. Opus is better for complex tasks and reasoning. I find it better for creating a plan. The tasks are small and well defined so Sonnet 4.6 doesn't have a problem implementing it. I'm experimenting though.

@OrenMe @cadamsdev @GitHubCopilot @burkeholland @code So why not use Opus 4.6 for the subagents as well @cadamsdev? Or do you specifically prefer sonnet 4.6 for them?

The @GitHubCopilot CLI Autopilot mode is super powerful Basically it is a built in Ralph loop(shoutout @burkeholland) but much more efficient due to how @GitHubCopilot utilizes PRUs Soon it is coming to @code It will relentlessly try to iterate your goal until ready BTW - Using SQLite for task list is a powerful tool

@notyuldshah @OrenMe @GitHubCopilot @burkeholland @code I have 2 custom agents. 1. Planner - Primary agent that uses Opus 4.6 to create a plan and batches of tasks. Then delegates those tasks to the subagent. 2. Implementer - Subagent that uses Sonnet 4.6. This one only implements the tasks that the primary agent gave it.

@cadamsdev @OrenMe @GitHubCopilot @burkeholland @code Did you prompt it to use parallel agents or something? How are u making it not forget the tasks it is doing?

@notyuldshah @OrenMe @GitHubCopilot @burkeholland @code Yeah looks like it does. I ran it for 8 hours and only took 2 hours 24 minutes to migrate the Angular app to React. It used 6 premium requests. That's not bad though thought it would be way more.

@cadamsdev @OrenMe @GitHubCopilot @burkeholland @code Does it count as more than 1 premium request?

@BenjDicken Create your own benchmark Ben or change PlanetScale to say "The world’s 2nd fastest and 2nd most scalable cloud database" 😂

Is spacetime cool tech? Yes Are these good benchmarks? No I've done a lot of database benchmarking. It's fun. I like benchmarks! But these were poorly conducted. - So many missing details. What cache warming was done? What region was each database and client in? What were the underlying instance types? Are we benchmarking databases or network latencies? - Anyone create a narrow benchmark that makes a particular database look good. It's better to use widely-used standards (tpcc, sysbench workloads, etc) - AFAICT the spacetime requests are heavily pipelined vs the Postgres options. - Why does spacetime get a custom rust client (rust is fast!) and all the others have to run slow js clients with an ORM middleman over http/rpc? Ultimately it's a big case of apples-to-oranges. Again, spacetime seems like cool tech! But comparing it to a 3-node HA semi-sync pg cluster is... an interesting choice.

@ladybirdbrowser Hope you don’t mind waiting hours for it to compile.

Ladybird adopts Rust, with help from AI ladybird.org/posts/adopting…

@youyuxi Have you tried OpenClaw yet? I hooked it up to telegram, installed the gh cli agent skill. Now I can say go create a PR in X repo from my phone. It's crazy. You can also setup a cronjob to rerun a prompt so it's building an app while I'm sleeping 😂

I haven’t been doing this a lot until lately, but AI can be such a good design partner. I have a Claude project with all the context of my current project, and I can chat with Opus 4.6 about it anytime on my phone. In the past I always capture inspirations in a backlog in Google Keep so I can do proper research on it later. Now I just chat with Claude whenever the bulb lights up. It doesn’t always come up with the best idea on its own, but it’s able to conduct research very fast, explore a wide variety of options, and list the tradeoffs to give me the necessary information to nudge it in the right direction - and honestly, that’s probably the only thing I am truly good at: intuition and taste on API design. AI amplifies my impact in the parts where I really matter and gives me a 10x leverage on the final results.

@youyuxi Integrating Vite+ with Node SEA?

I haven't been this excited building a thing for a VERY long time. Maybe even more excitement than when I was building Vite.

@youyuxi @jarredsumner Nice. Just out of curiosity how does Bun and Node SEA compare to Deno? Would be interesting to see.

Update: bun 1.3.9 improves a lot over 1.3.6 with bytecode on. It now has 25% faster startup time than Node SEA + code cache. Benchmark has been updated at github.com/yyx990803/bun-… Great job @jarredsumner !

@samuel_spitz replit app, "Rate exceeded.", building AI slop to detect AI slop on man... 😂

Suspicious that a website was vibecoded? Now you don't have to guess. Introducing ai-slop-detector.replit.app

@rough__sea Hahah looks like @deno_land agrees with you too @yarnpkg . x.com/yarnpkg/status…

just fyi if you're hitting segfaults with claude code, you can run it with deno and also get the benefit of less memory usage

@youyuxi Awesome feature. I'm gonna have to try this for my tui apps. I wonder if the executable size will be smaller because my Bun tui apps are bloated. 😭

In the next version of tsdown: it will be able to bundle your TS codebase into a Node.js SEA (Single Executable Application) binary with one command. And contrary to common belief, Rolldown-bundled Node.js SEAs startup is faster than bun --compile in my benchmark: github.com/yyx990803/bun-…

@pierceboggan @burkeholland Nice. Another thing to note is that "#" shows tools but CMD + / doesn't. Personally I like the CMD + / view more because I never use tools. Would be nice to make "@" be an alias for CMD + / instead 🙂 or provide an option that allows us to hide tools from the search.

@cadamsdev @burkeholland I agree, it’s silly and feels like being different just to be different PR: github.com/microsoft/vsco…

The Copilot CLI is…so good. I cannot recommend strongly enough that you start using it immediately.