O-1 requires yearly assessment of the exceptional status though. You can hardly plan your life around a visa that is quite subjective in itself and may depend on the mood of the USCIS officer reviewing your case on that particular day.
Ehh, the great firewall of China has been in place for a long time now, it's also illegal to evade it, and yet, it has sprouted many great tools that make it so simple to evade, that enforcement becomes impossible.
If anything, technology will always win over the legislation if it happens at scale. It may even lead to some new breakthroughs.
> Not to mention the big cloud providers are unhinged with their egress pricing.
I always wonder why this status quo persisted even after Cloudflare. Their pricing is indeed so unhinged, that they're not even in consideration for me for things where egress is a variable.
Why is egress seemingly free for Cloudflare or Hetzner but feels like they launch spaceships at AWS and GCP every time you send a data packet to the outside world?
They are just greedy. And they know nobody can compete with them for availability in every country. Except for Cloudflare, which is why it is so popular.
And god forbid you live in an authoritarian country and must use VPN to protect your freedom. Internet becomes captcha hell run by 2-3 companies.
I've had far fewer issues with my own bots that access cloudflare protected websites, than during my regular browsing with privacy respecting browsers and a VPN.
As a side note: I'm at least thankful Microsoft isn't behind web gatekeeping. Try and solve any microsoft captcha behind a VPN - its like writing a thesis, you gotta dedicate like 5 minutes, full attention.
Curiously, I explicitly tell all my LLM agents to never touch Pydantic models or environment stuff - it’s even in big, uppercase, bold text in my custom instructions for Roo-Code. LLMs seem to trip a lot over Pydantic’s magic.
pydantic-AI is a bit different than pydantic. The LLM isn't prompted to generate the pydantic model, instead it's encouraged to take input in the form of one model and produce output in the form of another.
LLMs (Sonnet, Gemini from what I tested) tend to “fix” failing tests by either removing them outright or tweaking the assertions just enough to make them pass. The opposite happens too - sometimes they change the actual logic when what really needs updating is the test.
In short, LLMs often get confused about where the problem lies: the code under test or the test itself. And no amount of context engineering seems to solve that.
I think in part the issue is that the LLM does not have enough context. The difference between a bug in the test or a bug in the implementation is purely based on the requirements which are often not in the source code and stored somewhere else(ticket system, documentation platform).
Without providing the actual feature requirements to the LLM(or the developer) it is impossible to determine which is wrong.
Which is why I think it is also sort of stupid by having the LLM generate tests by just giving it access to the implementation. That is at best testing the implementation as it is, but tests should be based on the requirements.
Oh, absolutely, context matters a lot. But the thing is, they still fail even with solid context.
Before I let an agent touch code, I spell out the issue/feature and have it write two markdown files - strategy.md and progress.md (with the execution order of changes) inside a feat_{id} directory. Once I’m happy with those, I wipe the context and start fresh: feed it the original feature definition + the docs, then tell it to implement by pulling in the right source code context. So by the time any code gets touched, there’s already ~80k tokens in play. And yet, the same confusion frequently happens.
Even if I flat out say “the issue is in the test/logic,”, even if I point out _exactly_ what the issue is, it just apologizes and loops.
At that point I stop it, make it record the failure in the markdown doc, reset context, and let it reload the feature plus the previous agent’s failure. Occasionally that works, but usually once it’s in that state, I have to step in and do it myself.
