bytearray

I understand the sentiment and I've been involved in software engineering in various roles for the last 25+ years. The thing that gives me hope is that never once in that time has the problem ever been that we didn't have more work to do.

It's not like all of a sudden I'm working 2-3 hours a day. I'm just getting a lot more done.

Nostalgia is a heck of a drug. :-)

Happy Thanksgiving!

Last week OpenAI pushed out an AI agent orchestration framework called Swarm. I liked how lightweight it is compared to other open source LLM frameworks that I find overly complex. Unfortunately, OpenAI positioned it as an experimental framework and on their GitHub page said they wouldn’t be accepting any PRs for enhancements. It doesn’t sound like they’re going to maintain it. That’s too bad because it is a nice approach for developers looking for a simple solution to orchestrating AI agents.

So I forked it and I spent some time making it work with Anthropic and Groq LLMs. The result is Microagent.

What are we appreciating exactly?

I would much rather they shut down.

Looks like Oracle employees have managed to beat the brigading protection on HN.

Same.

Do you have a link to the notebook?

Everyone should use the terminology I prefer.

Um, I think you mean 1337

I had a similar realization when I was at Google. Someone made the observation that if Google had a one-in-a-million error, that meant it was happening 50-60 times per second.

What happens when the last coin is mined and there's no incentive for people to run the network of computers that validates transactions? It seems like at that point the value would go to zero.

First off, thanks for doing something cool and releasing it as open source! Second, think the person 2 messages up has a valid point and I would like to offer the same constructive feedback.

As someone who is doing a lot of React/Tailwind development lately the more you can assemble larger pages from independently usable components the better. For example here: https://easyfrontend.com/components/ui/all/html?page=1

It would be useful sometimes to be able to search for a "card" component that I want to use rather than remember that I want to see cards I need to go to Blog #10. This seems to be my normal intent when I'm using a framework is that I'm trying to find a [card, dialog, select box, data table, etc].

Anyway though, don't let the feedback get you down, it's awesome and greatly appreciated that you're releasing nicely designed Tailwind UI components. Github repo starred and thread upvoted!

+1 to this sentiment. You'll also run into similar problems if you need to request quota increases. For companies that AWS takes seriously they'll approve the requests in a few minutes. For startups they'll literally take days reviewing the request before some random person at AWS decides whether to increase your limit or not.

Insurance companies are just legal scams. They take your premiums then come up with whatever excuses possible to avoid or delay paying out on legitimate claims.

bumpyball dot io is fun.

Interesting article, now time for me to go do more coding on the weekend. :-)

And this is why I don't have a lot of faith in leaderboards to predict real world performance. Every time I see something like this I decide to give Bard another try and every time it disappoints. Ok Google, I'm ready to be hurt again.

Cloud is the new mainframe, Kubernetes is the new job manager, microservices are the new COBOL.

Hi, I commented on the github issue, but maybe no zsh support is the problem. When I `echo $SHELL` I get back `/bin/zsh` in Wave.

I think I ran into this same bug: https://github.com/wavetermdev/waveterm/issues/161

I try to cd and it just leaves me in ~.

Using on mac.

Looks cool, but obviously pretty limited use when you can't cd. :-)

Marketing spam

"maternal love isn't always a curse".

What an odd thing to say.

I always thought Enceladus was just an icy moon, but turns out it's cooler than that.

Yeah, as a company they pretty much dominate the NoSQL space. 1B+/year in revenue and that market is still growing at like 30ish% YoY or so.

I was thinking more access to birth control and better education worldwide, the things that have historically have lowered birth rates.

The only real solution is fewer humans.

How does the performance of Causal Trees compare to other CRDT implementations, especially in scenarios with a high frequency of concurrent updates? It seems like a promising approach for collaborative text apps, but I'm curious about its scalability and real-world performance.

What strategies did you consider or implement to attract more users, and what would you do differently now to ensure better user acquisition?

At Google we used to have a wifi network you could connect to that was intentionally low quality for just this.