weli

I see reasons to have smart features in a fridge. Reminding you what food is going to be bad soon. Letting you know a list of recipes you can do with what food you have. Hell even being able to see and set timers without having to take my phone out. What I cannot fathom is the fact that the module running those smart features is not separated and modular enough so that basic functions like refrigerating are not dependent on an entire Android stack running.

That's what I don't understand. They could mitigate their name so much if they just split free/paid/enterprise. It's already shown that enterprise is much more estable and is largely unaffected from service disruptions. Why don't they go one more layer? For sure it's worth the extra complexity.

They affect different things. You want more powerful models -> more memory. You want the models that run to be faster -> more gpu cores.

Segments of the UI that don't conform to any other existing established design or conventions

Anecdotal, opinion:

Gpt is really good in vision stuff, or at least their MoE seems to be really cohesive. From my experience Claude models can be really good at language but the moment they need to look at a picture and decide why the design is not good what parts need improvement it degrades a lot. My easiest benchmark is giving them a screenshot of a feature in my app and tell it "identify non-normative UI blocks and improve readability and consistency". Sol does a great job at re-structuring the page into composable units that build upon each other and the general looks and feels of the app. Claude tends to over-focus one one part while completely forgetting about the rest or the cohesion as a whole.

You can compile to Linux targets though

A game inspired by living in Berlin for 5 years

Compactions are supported and heavily recommended! https://github.com/welidev/scroll/issues/1

Doesn't webhooks put even more work on them? You need to create a webhook dashboard, sometimes with multiple environments and administration endpoints, a way to set HMAC secrets, retriability and deliverability mechanisms, a local tunnel CLI for development, etc. With scroll if you already own a event log system the only work you need to do is exposing it to the client securely, same API keys as the rest of your existing REST api's.

> That seems unlikely to me, but I might be surprised.

I'm pretty sure you've hit the nail on the head. Especially now with LLM's you'd be surprised of how many people are running full on production apps that are coded solely by an agent, deployed with a mix of random providers (supabase + fly.io + whatever) and have no clue about what Kafka is. They are never going to ask for it because... they don't know they can, and why they should. They ask an agent "connect to stripe" and the agent codes your typical webhook ingestion mechanism, never proposes advocating for an event polling loop. Whenever there is a bug they will tell the agent to fix it and the agent will just write a webhook deduplication mechanism and that's it. You'd be surprised how little supposedly technical people care about the current state of technologies.

PS: Thanks for the feedback on the writing.

Not only this but the spec explictly allows provider-terminated connections followed by a 429 with a respected "Retry-After" if you want to definitely kick "slow" feeds. This combined with HTTP2/3 multiplexing makes holding feeds open pretty cheap.

I reached out through email :)

Just one correction. My spec doesn't force /scroll/ URL's, just proposes it as a convention.

Stripe events API is one of the examples of how to do things properly. And SCROLL is just trying to create a common spec so that everyone offers a stripe-like event polling api.

That is completely fine. In the spec cursors just need to be lexicographically comparable. They could easily be timestamps if the provider chooses to. In many of my example they are just alphanumeric dictionaries.

Maybe the LLM has written maximum 50 words of the article by just being directed to switch things around and improve grammar and or internal consistency

For the morale of the person who created the PR, closing every PR you make is pretty similar to just being banned

I'm not upset because LLM disclosure is mandatory. I'm upset about the consequences this has and how easy this makes for people to gatekeep new IC's under false and difficult to prove pretenses.

"This looks like AI, closed". And you as a new IC with no connections, no community visibility have no recourse to fight back against it. Anti-ai people love boasting about humans > systems and forget about those humans that will surely be negatively affected by this, even if low number.

If you are a new programmer who is just learning and spend time creating a PR for a project you have a lot of respect for and the first interaction is someone falsely accuses you of using LLM's then closes your PR you've now removed someone who could have genuinely grown into your community from ever joining.

Not only that but since now you are by default a suspect of having used an LLM (and you need to prove you didn't use one) why even bother not using one? You are going to get accused of it either way.

Please! Would love HN meetup in Berlin.

I really dislike every kind of rule that says something along the lines "if it's LLM generated you must self-disclose otherwise it is against the rules". Because this quickly leads to witch hunting and easy persecution of IC's under the pretense "it looks generated by an LLM" or they just investigate you and put you on cooldown from being able to contribute while they check for signs of "LLM-generated code". It's no different than laws giving any officer the right to violate your rights based on flaky probable cause "you have slurred speech therefore I will search your car to see any open containers". This can lead to a soft coup by anti-llm individuals which will make communities slowly drift towards a group of elitists that gatekeep any newcomer.

A simple mecanism:

Open an https stream to a collection, each new line is a new object event with a cursor.

Your server was down? Simply subscribe to the same collection from the last cursor saved.

Yea but that's where "reasonable suspicion" comes in a court of law. If you are standing next to a dead body, with a bloody knife in your hand.

Isn't english known for not having an official set of gramatical rules? There are conventions and rules from universities i.e Cambridge vs Oxford comma. But there is no single body that is commonly accepted by all english-speaking countries to be codifying the official rules of the language.

Maybe it has something to do with Cloudfront having an outage too https://health.aws.amazon.com/health/status

I guess the v8 isolate is heavily restricted and sandboxed and can't be used to access the local filesystem

At least with bun you can compile with bytecode enabled. That, tree shaking and other optimizations our production app entire docker image is like 120mb...

But slow organic growth is not a hyperscaling unicorn. Who wants to invest money for a mere 50% return? We need to x100 or x1000 it for it to even be worth to invest.

Not a situation that happened to me per se. More of a mix of situations that have happened to me and I've seen happen to people close to me aggregated into a single metaphor.

Really? Didn't use an llm other than to do a quick grammar check because english is not my first language.

If someone has the answer I'd like to know as well. I think the most important question to ask yourself is: Where did the story go sideways? At what point what character could have prevented the disaster?

For me there is no right answer. Maybe the engineer should have been more pushy with what things not to add. Maybe the founder entrepreneur should have been realistic. Maybe sales should have not had to promise things that were not developed yet. But to each of those there is a counter-argument of why that needed to be done in that moment.

Take it as a mental exercise.

I've been experimenting with writing longer-form content. I do agree the main point could be condensed a lot and I'm not the a great writer by far. This is kind of a rant and really cathartic for me to write after working more than 5 years on startups. Just wanted to share it.