figmert

The reason I use plan then implement is because I can adjust the plan, whereas if I get it to implement straight away, it might (and often does) make the wrong decisions that will be harder to adjust, or I'd have to adjust it after the fact.

For what it's worth I was joking. I've only been Shinjuku as a tourist, so I unfortunately won't be able to familiarise myself with the layout, but I do feel your pain. I've seen the same thing in smaller stations in the London and absolutely hate it.

Have you seen parallel?

Now we just need an app that links your location in 3d space so you can finally easily navigate Shinjuku.

Heavy disagree. I can't be sure about other AI coding agents as I haven't used others enough, but especially the items around prompt cache misses and compaction have hit me personally.

This article (combined with another article previously comparing CC and OC's system prompt, to which someone replied pi.dev sends even less) has made me want to switch agents, so I'm likely going to try out pi.dev and aim to migrate.

I hope so, but I am not holding my breath. OpenAI and Anthropic have too much money riding on them. I'll be surprised if they die.

Meanwhile I can't even get Fable to help me root my ecovacs robot vacuum :(

The website is down, but another commenter linked the original article[0]. Seems their reasoning is:

For each runtime, the fastest available web framework was selected to ensure optimal performance

But I agree. Best thing to do is use the same framework.

[0] https://hackernoon.com/myth-vs-reality-real-world-runtime-pe...

Whether you like it or not, software is inherently political. Most aspects of our lives are. The fact you're using GLM there's some underlying political reason (yes, my guess in this case is cost, which is also political) and mentioning it's Chinese backed is equally political. You may not care, but it's political.

When a model calls itself mechahitler and says there is a white genocide when there so obviously isn't, of course it becomes political. If we are not okay with supporting fascists, then in my opinion, we as humans are obliged to call out the issues in the same breath as the technical issues/advancements.

Well then the question becomes, what did it get trained on for next token prediction to come to the conclusion that there's a white genocide? Cos yes, people on the internet believe such stupid thing, but I can promise you, it is not the majority who believe that.

I'm going to try and solve it. Will share all.

Will he share all or will he try to sell you some fad instead? I wish him the best, and hope he recovers, but my money is on him trying to sell something new that won't work.

This has been my experience too. And that conclusion, I've found that if I really don't want to deal with such quirks, I can indeed literally just run the same docker compose file in rootful mode, and it works.

Making the changes required to run under rootless is often very simple.

Because Apple is not just a hardware company anymore. They track users and they sell ads. Sure, they are not at the same level as Meta and Google, but their ad platform is not insignificant anymore. Also that same software platform allows to get more money out of their users via their App Store.

Selling hardware with the software that helps them track means more revenue than the same hardware with the software.

I am still setting it up, but I've seen Mailrise and Apprise that essentially achieves this.

The nice thing about Apprise is that it can direct notifications to various places, not just another email server.

When I was working at Tesco some 7 years ago, the goal was to move to AWS. Surprised they're still not even close.

Good on them for sticking it to Broadcom.

Antibot measure also block real users at the slightest change they don't like. Anti-fingerprinting measure? You're a bot. Adblockers? You're a bot.

- No death by tapping y. You work on an isolated copy of your code in a fairly safe environment always so it becomes easy to get out of the way and just let the agent do its thing without potentially compromising your machine. Agents spawned by agentspace run containerized in yolo mode by default.

- Docker support inside the container via a docker-in-docker sidecar (Please note: `--docker` uses `--privileged` under the hood (required for DinD), so it's not a sandbox against actively hostile code. Use it for repos you'd trust on your machine anyway.)

These statements are pretty contradictory. If dind has privileged access, then that means the agent has essentially root access (or access of the host user if rootless)

Except when you have internet of course.

Yes I know you can run offline models, but it's hard to pass up on a little bit of snark.

This of course only supports http, not https. It's great for health checks e.g. in a docker environment. To do https, you'd have to use something like socat, but of course that doesn't use bash only.

I don't understand what's wrong with someone suggesting a book to an author? Do you think all authors have read all other books?

If you had pointed out the original commenter's patronizing comment, as if they with 100% certainty know better than the author who has just written a book about said topic (at least the commenter thinks so), then I'd have agreed with you.

Absolutely not. Maybe runtime overheads are minimal, but builds are so much harder to do. And yes, you need to figure it out maybe once, but it is still a lot more effort than just pulling in a new dependency. Now repeat that same effort for every new application, vs pulling that into every new application.

Profitable isn’t related to “worth the dollars”. You need to look at income and how much is being reinvested into growth. Amazon famously remained unprofitable due to reinvestment and waiting for them to become profitable before investing was a bad bet.

Sure, but we the only thing we know about the company is the current S1 filing. Need to time to see what all of that looks like. Fast tracking it and essentially forcing other people to buy without scrutinizing is the problem. They may very well be worth the money they claim, but we won't know until after they've proven it. That's what the rules are there for.

Why is that relevant? The rules are in place for a reason, why does it matter what the percentage is? They're not profitable. When they prove they're worth the dollars, they can be included, per the rules.

Also, S&P500 has a current market cap of $67 trillion, 0.3% of that is some $200billion. That is essentially a wealth transfer to the rich. They don't need it.

But why? Mise does this and more. It can install binaries from github, gitlab, uv, npm, and many more.

Not really my point. My point is more that you suggested no one has thought about this, but yes, they have.

To answer your question, there have been plenty of business who have created and published free software (albeit plenty have later closed them). Notable examples are Databricks, Hashicorp, Mongodb, RedHat.

Sure they've built a moat on top of their free software, but they have (or had) free software regardless.

As the article mentions, these arguments are basically all the arguments of the FSF, and everything Richard Stallman pushed for since the 80s. So yes, there has been plenty of thought, scrutiny, improvements, etc. 40 years of it in fact.

There's no reward for loyalty any more, and it's caused everyone to job hop (at least while that was possible), including me. At the time, employees complained about it, and in the same breath refused to give out any promotions and/or reward employees. Or they'd reward them with some shitty voucher.

The world has literally become the people vs corporations. There is no soul in working any more.

I don't know about it being dead, but i certainly stopped using it because it kills the context. A huge amount of tokens are wasted to mcp when in use. Skills use far fewer tokens. From my experience anyway. I'm also not advanced enough so maybe I'm not using it right.

Pretty sure there has been various efforts to do this, but they all eventually stall due to being unable to keep up with the GIMP source. I hope this time they're taking an approach that will allow them to keep up with the maintenance.

What do you consider a power user? Because I'd consider an OS that refuses to let you do what you want, and constantly reverts your customisation, the opposite of power user friendly.

We're a long way not because Linux cannot do it. We're a long way because publishers refuse to take it serious.