lvh

70 ms·

751.
by lvh 8y ago
They use a lot of similar techniques. One big difference is that docker uses user namespaces and flatpak does not. I'n not sure about the reasoning, but it's probably a combo of "not trusting user namespaces" (disagree)
752.
by lvh 8y ago
We're not. But: good news: people are experimenting with verifying specs at compile time! And since specs come with quickcheck built in, you get a lot of "free" (in developer hours) "compile time" (actually: develop
753.
by lvh 8y ago
> In this sense Java is actually dynamically typed: there's no way of knowing at compile time whether an object is actually an instance of that object or is null. That is an extremely unusual definition of dynamic typing. If I have
754.
by lvh 8y ago
Sorta! Cheshire gives you a key-fn entry point. For small JSON objects you may just want the same namespace and then that is sufficient. Otherwise you can look up the right namespace in that fn or walk that tree (I’ve used specter for this)
755.
by lvh 8y ago
I can’t parse that second paragraph. Can you elaborate? Runtime specs absolutely also validate. And you can run them at test time. And you can put arbitrary code in them if you want. And sometimes you can statically validate them. What I’m
756.
by lvh 8y ago
The author is one of the principal folks behind Haskell. But I can’t help but point out that I promise you you’ve never heard of the five or so lanaguages he mentioned as an alternative approach, and somehow this Lisp thing is still around.
757.
by lvh 8y ago
Clearly languages have common idioms. Lisp just has most of them implemented in terms of very basic idioms. Saying “macros are bad because you can overdo it” is a little silly. Sure: prefer plain old data manipulation over syntax manipulati
758.
by lvh 8y ago
Thank $DEITY that impracticality is always because you’re not using the One True Static Typing School! Seriously: you’re describing a feature that only exists in a small subset of languages and only a small subset of contexts at that. OCaml
759.
by lvh 8y ago
It's hard for me to provide a good example without knowing some of the stuff you're in to but here's one of my favorite macros: https://github.com/lvh/caesium/blob/master/src/caesium&#
760.
by lvh 8y ago
Sibling comment provides a great resource, but I think a good place to start is explaining macros. At the end of the day, every line of code you write is some data in a bespoke data format. Most programming language implementations leverage
761.
by lvh 8y ago
Function specs are not about compile time verification: they are about checking that the object you got is the thing you thought you had. Compile time verification is nice if you know all your inputs at compile time. At some point, data fro
762.
by lvh 8y ago
Do you feel like the process described in this post is a facsimile of ITIL? Do you feel like ITIL is generally associated with repeated shipped product?
763.
by lvh 8y ago
I would like to know more about the review methodology and when the review is triggered. If I ask for wide permissions, get them, and then sell my extension to a malicious third party, what triggers the review?
764.
by lvh 8y ago
There are good tidbits in that post, but the way you measure "smart" is awful and almost entirely subjective. The conversation "just flows"? Really? Gee, I wonder if that's likely to make you hire people who have th
765.
by lvh 8y ago
There's a critical kernel of truth here that I want to point out: work-sample tests. The longer I spend building teams, the more I am convinced: there are people who do WSTs as a way to qualify applicants, and there are people who just
766.
by lvh 8y ago
Matasano used to bet the farm on it, I did it at my last job, and Latacora's doing it again. I don't think what those companies do is trivial. (Disclaimer: I'm a Principal at Latacora.) In my experience, when you think you ca
767.
by lvh 8y ago
You could, instead of relying on a resume to determine if someone can do X, ask them to do some X instead of an interview, and then use an objective rubric to decide if they did X to some level of satisfaction. This is called work sample te
768.
by lvh 8y ago
OK, so there's an attacker on the network in both cases (UXSS and the worst-case-dnsmasq-vuln). So, to compare the two, you look at what else you can do -- and UXSS clearly wins there. "It wouldn't be hard to mount a phishing
769.
by lvh 8y ago
I did not say "a compromised DNS server is completely inconsequential", I said that a compromised WebExtension with :// / and tabs permissions has UXSS (obviously true) and UXSS is worse than compromising DNS resol
770.
by lvh 8y ago
You can't dockerize kernel modules, no. Though Linux mainline wifi support is pretty great these days, coming from someone who vividly remembers using wpa_supplicant with Windows drivers :-)
771.
by lvh 8y ago
This is great: it has most of the benefits of an extension without the concern the extension gets compromised (you already have to trust your browser). The only downside is it doesn't address ads in things that aren't browsers, li
772.
by lvh 8y ago
You can run argo-tunnel/cloudflared on it and use that. You'd still be taking plain-old-DNS in, but the arguments in favor of DNS-over-HTTPS aren't as strong on a network you control. I don't think anyone has written cus
773.
by lvh 8y ago
You don't have to be confident has "no vulnerabilities" (an absurd standard) to understand that the worst possible vuln in the DNS server (say CSRFable RCE in dnsmasq) still puts an attacker in a less privileged position than
774.
by lvh 8y ago
I'm hearing confirmations that it's just Yubico, yes.
775.
by lvh 8y ago
Yes, they do. WebAuthn is the modern API, but it’s backwarss compatible with every U2F device. You don’t need a WebAuthn key to use the WebAuthn API.
776.
by lvh 8y ago
This is underwhelming. - Requiring the attestation cert and not accepting a self-signed one—which U2F devices will magically not work? I don’t know, but apparently it’s strictly Yubikeys now. - Still not convinced anyone seriously uses the
777.
by lvh 8y ago
Do you care if the auth is GPG or U2F? Seems like you want -L to forward a socket, and GPG_AGENT_INFO to point to it. I tweeted this series a while ago because it’s hard to find how that works: https://twitter.com/lvh/s
778.
by lvh 8y ago
My ISP doesn’t have anywhere near as shady marketing practices. Does yours regularly ignore do-not-call and do-not-mail requests? Do they send misleading envelopes that don’t have their brand on the outside? Mine (AT&T) doesn’t.
779.
by lvh 8y ago
Dang, TIL; I could've sworn I've made that work on a NEO. PIV mode presumably also has forced touch/reauth for signing?
780.
by lvh 8y ago
I don't understand what you're saying. Which resident keys? WebAuthn adds a number of crypto schemes -- to wit, I think they add RSA. You can certainly deterministically generate RSA keys but it's a lot more of a pain in the

More ›