I don't think this is more challenging than the browser thing. The scope is much smaller. The fact that this is "only" 100k lines is evidence for this. But, it's still very impressive.
I think this is Anthropic seeing the Cursor guy's bullshit and saying "but, we need to show people that the AI _can actually_ do very impressive shit as long as you pick a more sensible goal"
From the docs I think Limmat is much more minimal. It doesn't have a merge queue or anything, "jobs" are just commands that run in a worktree.
I would be interested to try SelfCI coz I have actually gone back and forth on whether I want that merge queue feature in Limmat. Sometimes I think for that feature I no longer want it to be a local tool but actually I just want a "proper CI system" that isn't a huge headache to configure.
I was a bit skeptical at first coz my experience so far has been that Nix's weak typing quite rarely causes me problems.
(Likely because NixOS options tend to be how I communicate between pieces of code, and those have a good enough type system).
But actually I think, as something that you just use occasionally like in your JSON example, this could be really cool. I wouldn't wanna use it as "gradual typing" in the sense of aiming to gradually type a whole body of code I think I would be more interested in "selective typing".
Just occasionally when you get that itch of "ugh this will fail really ungracefully if the inputs are bad" then you just apply this where it's needed.
Getting Google Docs to be a Word alternative was an order of magnitude easier than getting GCP to be an AWS competitor.
Now that AWS has two serious competitors (and some non serious ones), privately funding another one just seems impossible to me. Who is gonna chip in tens of billions of dollars to fund "that, but European, and 15 years from now"?
I think the only ways we can get serious Euroclouds is some combination of:
1. EU intervention (nasty regulations and expensive subsidies).
2. People using non-equivalent products (Europeans have to use lower-level infra and do a lot more ops in-house). This part would have its upsides anyway TBH.
> Just mandate EU countries' public administration to rely exclusively on EU cloud solutions.
This happens already in some areas and it is not cheaper or better. The EU funds national clouds where public institutions use them. What does it mean? VMware with Tanzu or OpenStack. And then some services thrown in to offer some S3 like buckets and that's it. The rest has to be built by the beneficiaries. Servers? Brand names like Lenovo/HP/Dell. Storage? Brand names like NetApp, HP, Dell, Lenovo, 3Par, IBM and the list goes on. Networking? Cisco (mostly), HP/Juniper. Firewalls? Cisco/Fortinet/PaloAlto/CheckPoint/etc.
Basically an enterprise setup masquerading as a cloud offering.
And even if there would be EU wide offerings for such cloud, there's too much money at stake to let institutions from one country buy services from another.
Yeah. The EuroCloud will always be dramatically more expensive and much much worse. Anyone who's claiming otherwise is living a fantasy. The only argument that makes sense is "but it's worth it".
(One detail: it will be much worse at the margins where current clouds actually compete. But actually I suspect only a small number of our customers actually exist at that margin. I think a lot of people are just massively overpaying for their cloud platform and so they might be fine with a EuroCloud anyway. This is why you hear stories today like "we switched to Hetzner, halved our bill, and it works exactly as well as the AWS products we used to use).
> And then some services thrown in to offer some S3 like buckets and that's it. The rest has to be built by the beneficiaries.
Ditto, a fully featured EuroCloud is not gonna happen. Again, it has to be worth this cost.
> Brand names like NetApp, HP, Dell, Lenovo, 3Par, IBM and the list goes on.
This is the only part where I disagree. I think it's OK if the EuroCloud is built out of US hardware (like how the AmeriCloud is very far from free of Chinese hardware). Obviously presents a significant risk re supply chain security but still, the _really_ important thing is actually sovereign operations. The most important thing is who has the keys to the DC.
> This happens already in some areas and it is not cheaper or better.
The goal is not for it to be cheaper or better. The goal is to have money spent on domestic actors. Will some of them provide a dogshit service? Sure. Just like it happened in the US. Time will sort things out.
> What does it mean? VMware with Tanzu or OpenStack. And then some services thrown in to offer some S3 like buckets and that's it. The rest has to be built by the beneficiaries. Servers? Brand names like Lenovo/HP/Dell. Storage? Brand names like NetApp, HP, Dell, Lenovo, 3Par, IBM and the list goes on. Networking? Cisco (mostly), HP/Juniper. Firewalls? Cisco/Fortinet/PaloAlto/CheckPoint/etc.
You need to cut this purity bullshit where Europe must own all the stack from the foundry, or do nothing. What you're building is an unwinnable battle.
I have a PiKVM attached to my PC at home, so at some point I'm thinking of setting up a crazy demand-scaling scheme where when my underpowered homelab nodes can power up the PC when they need to run a heavy workload.
WoL is easier if it works. My experience has been that with consumer hardware it usually doesn't. Debugging it is more hassle than it's worth IMO. I think if you don't have a proper mobo with a BMC then just throwing in a KVM is easier on average.
WoL is reliable when waking from sleep/suspend. I have yet to see consumer HW that can do it from poweroff. But if suspend is fine, all you have to do is configure your firmware to turn on after power loss (so you always boot into your OS) and in your OS enable WoL and configure suspend on inactivity as you wish. It should be reliable and failures would default to "on", not off.
I have lately taken to this approach when I raise bugs:
1. Fully human-written explanation of the issue with all the info I can add
2. As an attachment to the bug (not a PR), explicitly noted as such, an AI slop fix and a note that it makes my symptom go away.
I've been on the receiving end of one bug report in this format and I thought it was pretty helpful. Even though the AI fix was garbage, the fact that the patch made the bug go away was useful signal.
There's more to it than just coding Vs building though.
For a long time in my career now I've been in a situation where I'd be able to build more if I was willing to abstract myself and become a slide-merchant/coalition-builder. I don't want to do this though.
Yet, I'm still quite an enthusiastic vibe-coder.
I think it's less about coding Vs building and more about tolerance for abstraction and politics. And I don't think there are that many people who are so intolerant of abstraction that they won't let agents write a bunch of code for them.
In the specific case of __jump_table I would even guess there was some work in getting the Clang build working.
reply