HN Simulatornew | past | comments | lists | submit | turblety's commentslogin

Yeah it's infuriating. I love how even Apple are embarrassed by it, since in the first two animations of people holding and rotating the device they've left off the camera bump, making it look smooth flat.


The entire publicsuffix thing is such a mess. The fact that project even exists is ridiculous. Why can this not be a dns setting?


They should have never allowed cross subdomain access for cookies. Every subdomain should be treated as completely unrelated to all other subdomains.


sigh

See, before the cloud, you used to deploy your code to actual physical servers. If your site became really popular you needed more than one server to run it.

Before commercial and open source load balancers became a thing we literally used HTTP redirects to distribute load. You might visit www.example.com and get redirected to www13.example.com, which was an actual box in a datacenter.

I don't think the internet would exist today if websites just randomly fell over once the number of simultaneous users exceeded the capacity of a single server.


When I was at yahoo, we certainly had load balancers, but we didn't have super fancy load balancers that could separate traffic out by path, so each 'property' would use a different hostname (i worked on travel.yahoo.com, which is dead these many years), but we all shared a login cookie (and some other cookies).

We'd have certainly adapted if we couldn't share cookies, but it would have been a lot messier.


Why? That looks like a mess. Proxies existed before http, so why not just proxy traffic between your various physical servers to manage load? It’s literally a poor man’s load balancer. It how Travelocity worked way back in the day.


Then use an http header like we do with CORS or something, which lets a domain set a cookie for another domain that opts in. Or, you can just include the necessary cookie info in the redirect.


Couldn't you just ghetto load balance by using internal DNS to round robin to a spread of servers with different internal IP ?


DNS caching and thus the load distribution is outside the control of the entity running the servers. You assume you get roughly same traffic on each of the published IP addresses, you even think you can advertise some address twice so that it gets twice the load vs other severs (if you just set up a server capable of serving twice the traffic), but the reality is one of these IP addresses might be used for 90% of the requests (as an example) due to caching and bad luck.

So no, you cannot reliably control load balancing using one domain name pointing to multiple IP addresses. (IP anycast is better suited, but much harder to implement)


One of the complications that gets discussed is that there are way too many use cases with different requirements on what it means for domains to be related. The cookie problem is something browsers need to solve, so its perhaps reasonable that Mozilla (and volunteers) manage a list that focuses on that user case. But that's poorly suited for other use cases.

There's a lot of people thinking about this though. One somewhat related example is https://datatracker.ietf.org/doc/draft-manelidis-dnsop-domai..., which has a double-opt-in DNS TXT record mechanism. That draft and forthcoming discussions on the mailing list are a good place to understand the complexities here.


If, for example, you could trick a client into thinking "com" was not a public suffix, then from malicious.com you could set cookies on paypal.com

DNSSEC is not widely deployed, and so we cannot rely on the authenticity of DNS.

Maybe if we had DNSSEC, then we could rely on it for publicsuffix. Maybe lots of others things could be more easily solved if we had that too.

DNSSEC automation is coming, e.g. https://www.rfc-editor.org/info/rfc10026/. It might never become mainstream, but don't let anyone tell you that we couldn't benefit from having DNS authenticity!


because, like everything, bad guys would abuse it, causing a list of "bad publicsuffix" to exist, which is basically the same thing.


Could we just publish (and cache!) TXT records at the appropriate-level domains? E.g.:

  host -t TXT github.io.
  etld_depth=2


badguy.com would set the same flag, allowing a.badguy.com, b.badguy.com, c.badguy.com to count as fully independent sites


(Vouched so I could reply.)

Why would that be a bad thing? What happens under badguy.com is badguy.com's business, is it not?

Is it any different than setting cookie scoping rules?


The cookie use case is just one use case. Another PSL use case is helping browsers decide what part of the FQDN is most important for the user to notice. On firefox I see "ycombinator.com" in bold, with normal font weight on the news subdomain. On smaller screens the URL bar isn't left aligned, it adjusts the view to prioritize showing the relevant domain name. You don't want an attacker to control how the URL is displayed in that security context.


It’s also strongly used by chrome and similar browsers so malware at randomuserdomain.geocities.com doesn’t block the entire root domain.


Yeah go for it. It was really fun to build and I learned a lot about how these harnesses work. Definitely recommend building something like this.

For me, the use case is in my Remote Terminal app I'm building [1].

I have an agents sidebar that shows every running agent on the machine with the state. So I end up creating a kanban like board for all the agents based on if it needs my attention, is done or still working.

1. https://terminay.com


Hey, yeah I have two rules:

1. I don't want to extend the the harnesses in any way, not even hooks/plugins/etc 2. I don't want to poll, ever. It must be system events, file system watches

So I have to basically watch each session file from every harness for changes and derive a state from that.

Most CLI's have a file per process with some stats like the pid, state, is idle or not, etc, which is really helpful.

Each agent has a small provider that knows how that specific harness works, and watches it for changes. e.g. ~/.claude/sessions/{pid}.json

Those registries are small and change much less than transcript formats do.

But yeah, I think this will always be a matter of keeping it up to date.

I would like to retain support for older harness versions as I go.


I'm not sure. The problem with the cloud llm's is they are complete black boxes that change frequently and randomly day by day.

If you run Qwen 3.8 on your own hardware, every single day, it's the exact same model running in the exact same way.

Yes, it's nowhere near as "smart" as the cloud based models. But it's consistent.

So the workflows and "ways of working" you create will work mostly similar day to day.

With Claude/OpenAI you frequently find days where the models are useless, and days when they are out of this world.

So I guess the choice comes down to:

1. Randomly the smartest thing on the planet with unpredictable rate limits that is mostly amazing, but frequently messes with your workflows

2. A really good local coding model that is consistent every day with no rate limits

I'm not sure. My gut feeling is maybe the right answer is a mix of both.

Gambling on the biggest models, hoping they are working smart that day, when planning or doing very complex work. Then doing most of the tasks/daily work using local models??


You can run any open model on a shared API host via OpenRouter and pin to which host you want to go for the quant/privacy/etc. mix you care about. You can pay them directly if you don't want the OpenRouter overhead - but the convenience of switching, having one invoice, etc. is worth it IMO

It's not closed hosted models vs open local models, it's hosted open models vs local open models where the math doesn't work for local LLMs.

The only local inference use-case I can think of is porn generation (because most providers don't want to deal with it) and illegal shit like hacking to minimize the tracing.

And if you're super paranoid - but honestly giving sensitive info to LLMs in any scenario is a gamble.

If you game and can use your GPU I guess then it works as well but models that fit into a gaming GPU suck too much to bother IMO.


I think the answer to this is yes, it'll be just like "reading and understanding your assembly". We just don't check what instructions your code compiles down to anymore, and I think the same will be for code.


100% agree. I've ended up building my own [1][2] mostly because I got annoyed the agent cli's are building remote control into the agents, instead of into the terminal. I really want the agent cli's to focus on being the best autonomous agents, then let me choose a great terminal that does all the extra.

Another great option (not mine) is Orca [3] and it's a lot more mature. Really nice to use.

1. https://github.com/markwylde/terminay

2. https://terminay.com

3. https://www.onorca.dev


I'm glad it works for you, but I've got enough electron apps running on my computer already. I have to run a bunch for work and fully half my cpu is just electron at idle


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: