Canadian software engineer living in Europe.
I’ve done some digging into this one, and despite the interesting side note that the main proponent of a recent datacentre project up there is named “Danny Quinn” (note my username) the best I can tell is that it hasn’t been built yet so I didn’t see anything about going back on a self-powered promise. If you see anything more on this front though, I’d very much like to see it!
It’s been years since I was experimenting, trying to set things up properly, but at the time I could swear that I’d have the following set:
PubkeyAuthentication yes
PasswordAuthentication no
UsePam yes
and attempts to SSH into the system would still result in a password prompt, even though PasswordAuthentication was set to no. After some experimentation, I found that if I set UsePam to no, you wouldn’t get a password prompt, so that’s what I’ve been doing on every SSH server I’ve had since like, 2000.
Tinkering with it now though, I see there’s also another directive, KbdInteractiveAuthentication, which when set to no, will ensure I don’t see a password prompt even when UsePam is set to yes. I can only assume that 26ish years ago there was no KbdInteractiveAuthentication, or I just didn’t know about / notice it!
Well no harm done in my case anyway, as I don’t generally expose SSH servers to the outside world anyway. But it’s good to know that I don’t have to turn off PAM to disable password logins anymore.
I’ve been setting UsePam = no for decades because leaving it on meant that it’d still present a password login of keys failed. Is this still the case, was I wrong from the beginning, or is this somehow ok?
I don’t know where you saw this “No”, but all of their code has been up on their (self hosted) GitLab for years: https://git.jami.net/savoirfairelinux
I believe the quote is:
“No one will ever need more than 640KB of RAM”. – Bill Gates
It just wasn’t clear what was being downloaded.
For example, here’s a screenshot of the UI when I tried to use it on this page (I was logged in at the time, which grants access to the full video):
The UI presents 3 options, all of which are basically identical, each with impossible file sizes. I ended up clicking [] on each of them 'til I found the full-length video. The preview opens up in a new tab, rather than inline or in a pop-up which wasn’t ideal (I had like 20 tabs open, now my extension tab was on the left and the preview for one of the streams was way over on the right).
It worked well though, once I had the above figured out. Here are my suggestions if you’re interested:
Regardless, it’s a great job, so thank you for sharing it! Now I can watch Zeteo videos on my own time without that useless Substack app!
I was dubious. “What could this possibly do” I thought, “that yt-dlp can’t?”.
I pointed it at a video on Substack that I’ve been wanting to watch offline (that platform is just shite) and after a little getting tripped up by the UI… boom. 1hr of video just in my Downloads/ folder.
Thanks. This is pretty great.
That’s alright, the market’s not really interested in bearing Anthropic’s costs either, so maybe this will all be a wash.
Epic will simply pay Anthropic a “sabotage your LLM to refuse to undermine our products” fee and then they’ll pass the costs onto their users.
The only useful LLM (and I recognise that that word is doing a lot of heavy lifting) is one that you own and control. Everything else is just letting some techbro limit what you can do.
He compared an LLM to a compiler suggesting that the concerns about using them are the same as suggesting that everyone should write in assembly.
The comparison is invalid, since one mode is deterministic, while the other is not.
LLM-driven development leads to programming via the abstraction of the model, disconnecting you from the source to the point where you don’t understand what it’s done or why. He’s suggesting that this isn’t a problem because an LLM is just like a compiler, which suggests that he’s arguing in bad faith or doesn’t understand the “tools” he’s using.
Does he… think LLMs are deterministic?
Your work lets you use Linux?
Honestly? Gentoo. It’s got a steep learning curve, but the docs are great and after your first install, you’ll know more about Linux than most users.
I still use the em dash, for precisely your reasons above!
Well if you do decide to attach it exclusively to domains, you might want to consider developing a DNS TXT standard format so this sort of information can be stored in DNS and cached accordingly.
and I don’t see any value in limiting classification to the domain level. How would one classify wikipedia.org in this scenario? Would it not make more sense to define an open standard that’d leverage this system but allow domain managers to define it themselves?
To take my university.tld as the example, that university might host a file called at https://university.tld/ocs.json that looks something like this:
{
"/": "EDU"
"/math": "MAT",
"/journalism": "JOU",
}
(Heads up to OP: there’s no journalism in your current spec. That feels like an oversight.)
This would allow the high-content site admins to classify parts of the site differently. The /ocs.json file might even be dynamically generated for complex sites like Wikipedia.
I’m curious as to why you’d limit this to the domain level. A university for example might have thousands of URLs in it with wildly different subjects:
university.tld/math/student-name/thesis-on-mathy-subject/university.tld/journalism/student-name/big-story-about-politicsHow would your system account for this?
The massive amount of fossil energy being wasted on this masterbatory nonsense is literally killing us, and we clearly deserve it.
Ooh, this is very helpful! Thank you!