Relatedly, never talk to me or my son again.
https://mailarchive.ietf.org/arch/msg/tls/IetbJONEoHH-o3bXSmHacPlIX4o/
Relatedly, never talk to me or my son again.
https://mailarchive.ietf.org/arch/msg/tls/IetbJONEoHH-o3bXSmHacPlIX4o/
Claude broke a PQC algorithm with classical techniques.
made available to the Zed people, so everything gets faster in a text editor.
it's always a text run cache thing, sigh
I guess we could just call it "Corporate Memphis" writing.
There should be a phrase for writing that resembles "Corporate Memphis" in graphic design.
Like, ok, here comes the pat conclusion. Whatever it is that you get in The Atlantic or the NYT.
Disingenuous Africa advocacy might be my least favorite thing.
Today: "we can't use TLS because Africa". Me: "It worked in Algiers and Addis Ababa, what are you looking for?"
@ekuber I have an interesting one. I wrote a library that was pretty fast with lots of unit tests. Then I told an LLM to rewrite it with well-scoped lifetimes and no unnecessary clones. That worked!
reminded that I have read Netscape 4 C++, and that made me wholeheartedly support Rust. It made me think "this should be illegal", and now it is.
TIL there's a Rust podcast: https://corrode.dev/podcast/s03e01-zed/
I wouldn't close the loop as of now. Claude suggests some pretty wacky things about 1 in 10 times, and also strangely gives up like a very junior engineer ("let's just do this..."). It seems tuned to get a task "done" if it hits difficult problems.
Huh, so I can run a Rust binary under samply, read the profile in Firefox or Chrome, and then tell Claude Code where I think there are bogus hotspots. It rewrote a bunch of Nom parser code to allocate as little as possible.