jtwaleson

Imho it's wildly ignorant to call the current SotA coding agents a slot machine. They are extremely powerful tools.

My point was: I play Factorio for fun and it reminds of me of what programming was like before AI coding. In Factorio, all the automations you build are proper abstractions that you yourself actually designed.

For pre AI software engineering yes.

There are many many instances where the CEO is not an employee, but they operate via their own legal entity. To be honest I don't know how this works at larger companies.

A couple years back I drove from Vegas to Yosemite. Due to a road closure we had to take a 4 hour detour. We had this colourful van from Escape Campervans, and every night looked at freecampsites.net to find a free place to sleep. Nice adventures.

That night around 8pm we drove to a random pick, called Willy's Hot Springs, very tired and hungry. We had no idea what it was and had never been to a hot springs. After confirming we could stay there, we drove for dinner and a beer at Mammoth. When we got back, we were curious and went along the wooden path to see the hot springs.

It was around midnight, there was no moon and 0 light pollution. It was magical. You sit in the cold night in natural hot water, in the middle of pretty barren country. There's all the stars and the milky way. Then all of a sudden it got pretty bright. The moon rose over the horizon. It was the first and only moonrise I've ever seen.

This was the most magical night in nature I've ever experienced. A friend of mine said "oh, Willy's Hot Springs, I've been there too, it was kind of mid." I think we experience the world in different ways.

yes

I think it's because people are building agentic flows, reducing the amount of developer seats needed. It's the first step towards usage based pricing.

I was young and naive

At one point, 20 years back, I was joking that The Netherlands should quit the EU and join the USA as the 51st state. Lot less appealing in the current political climate, but for technology it'd be great. My point was that the constitution of the US is based on ideas, and the country could be open to any territory, independent of location, whereas all of Europe's countries are based on more on historical artifacts.

Funny that the opposite approach is now considered with Canada. I can see it though.

It's a best-in-class real time collaboration board. It's also a decent general purpose diagramming tool, but if you're using it for creating software architecture diagrams or product flows, there are better tools for that.

I have some friends there. Morale in the company was very very low. There were multiple rounds of lay-offs and management went AI crazy. They were desperate to live up to their extreme valuation from 2022. "We have to be the future of work" etc. Now the exit was 10x lower than their max valuation.

Great whiteboarding tool though.

Can someone in Microsoft confirm my suspicion that this thread is still going and nothing was fixed?

Ingesting the whole codebase in their cloud and then maintaining indexes of where everything lives and what already exists. Not sure if that's where they are going but that's what I feel looking at some of the beta functionality and announcements.

For the forges, what the sibling said.

The tide is turning though. I was using the old non agent view in cursor every day for about a year when I built my startup with it. Then Claude Code blew it out of the water. But since two month or so I much prefer Codex and Cursor with the agent view.

However, what I see cooking at Cursor is way more promising than the others. Things like full system understanding, their own forge, multi-repo support. I'd put them as way more visionary in the Gartner magic Quadrant ;)

At Comper we have a very hot key-value store for annotating git data. We maintain a parallel git-blame data structure so we can do incremental "git blame -w -M -C -C". Typically a very expensive operation, but if you make it incremental, you can make it very cheap when new commits need to be analyzed. However, building the git blame tree is still pretty intensive for large repos.

We currently use rocksdb with storage on the same node, and hit rocksdb 1000s of times per second during our analysis. About 20% writes, 80% reads. The issue is that we need to start scaling horizontally, for burstable workers and zero-downtime deployment. So we're thinking to offload to an external kv service instead of a local rocksdb.

TiKV seems a good replacement, about 3-4x slower, but very scalable. Reading this article, I think a separate postgres cluster with unlogged tables might be a good idea. If anyone has some experience to share, let me know!

I think it's just to signal that prices will go up in the future.

I've been using Taskfile. Are just / mise much better?

Same approach here. Sometimes it's hard to debug. "Why is this weird test failing for an unrelated change", but I have to remind myself that it's a real concurrency bug that would have otherwise gone unnoticed. Worth the effort in my book!

*west -> east? ;)

They just cancelled a train from Amsterdam to Munich for me and never replied to my form for a refund. I called them about. Had to spend 600 extra for a sleeper instead.

IMHO you can remove the "good" and "except when". DB is 30% of the time a frustrating experience, which I've noticed makes me want to avoid it, even though I love trains. Trains in the rest of Western Europe are pretty great!

Very cool, this is the first language I've seen that has effects/permissions annotated.

When reasoning about a bit of code, types and effects are two things that help me grok it. However, there are a lot more. Like: how much memory can this use, does it process any PII, how quickly should it respond, which users are allowed to run it. Our "type" system could then alert us when we call a function that takes 20 seconds straight from a button without a loading spinner.

I'm very curious to see whether a language with many annotations this could work. Probably a language stored in a database approach like Unison is needed.

Other than that, I'm also curious to know if Jacquard works well with LLMs in practice. How well does providing the language docs into the prompt actually work? The models are trained on mainstream languages so I've been doubtful whether new languages will be able to compete.

Super happy user here! It's an excellent piece of engineering.

We're running a fork that supports a "sidecar" server for running multiple integration tests against. So if any tests that need the server are included, it spawns the server, runs the integration tests, and then shuts it down. By re-using the same server we speed up our runs tremendously.

Discussion thread on gh: https://github.com/nextest-rs/nextest/discussions/3330

Love your games Zach! Hope my kids will take to them at some point.

I'm showing this page to my team and investors every couple of weeks. Visual, animated explanations are MUCH better than textual content for deeply grokking something. This is what we're trying to build for large software systems. I love the animations on this site so much, thank you for building them.

I've been thinking that there might not be new programming languages. The amount of code in the current popular ones will explode, so that's what all LLMs will be trained on.

Good luck coming up with a new language and getting enough content out there that LLMs will be fluent with it.

If true, I think that's a shame. There's plenty of innovation still to be done.

Yes and yes. Uber for 260 eur arrived within 5 minutes after booking.

I was at a conference in Frankfurt, traveling back to Amsterdam with my cofounder and got stuck in Oberhausen. We have an early flight tomorrow and there's no trains in NL due to a strike tomorrow morning, so we decided to take an uber home.

At first the delay was 30 minutes. Then 2 hours. After 1h30 with zero updates we decided to bail. Just checked and nothing is moving yet, so we made the right call.

Nope, unfortunately not right now. Thank you!

we do a very comprehensive scan in multiple dimensions: git blame history, llm passes over every file and a tree-sitter analysis for cognitive complexity. We bring our opinionated approach for tools (do you have linters, CI/CD, static typing, agent docs) and score each aspect. We also create your architecture diagrams with our secret agent sauce, and map technical debt on each level of the C4 hierarchy.

bespoke canvas (based on pixijs) + sync engine. We want to be fully on-prem capable.