Danny Lin(@kdrag0n) 's Twitter Profileg
Danny Lin

@kdrag0n

building a better docker @OrbStack · dev exploring new things · @stanford

ID:929858958935834624

linkhttps://kdrag0n.dev/tw calendar_today12-11-2017 23:50:26

2,6K Tweets

13,8K Followers

589 Following

Danny Lin(@kdrag0n) 's Twitter Profile Photo

more fun:
- on macOS, SIGPIPE is always sent to the main thread regardless of which thread caused it
- so it's impossible for the Go runtime to implement this FFI behavior on macOS: pkg.go.dev/os/signal#hdr-…
- Linux has no SO_NOSIGPIPE, so every call needs send(MSG_NOSIGNAL)

account_circle
Danny Lin(@kdrag0n) 's Twitter Profile Photo

Unix surprise of the day: SIGPIPE

Writing to a broken pipe/socket causes SIGPIPE to be sent to the calling thread, killing the process by default??

I guess it makes sense for CLI programs...

Solution: everyone just ignores the signal with SIG_IGN or (macOS) SO_NOSIGPIPE

account_circle
Danny Lin(@kdrag0n) 's Twitter Profile Photo

New commit in xz's upstream repo to fix the sabotaged Landlock check (2 hours ago). Lasse Collin is probably clear?

New commit in xz's upstream repo to fix the sabotaged Landlock check (2 hours ago). Lasse Collin is probably clear?
account_circle
Danny Lin(@kdrag0n) 's Twitter Profile Photo

Been trying to downgrade xz with Nix:
- overlaying nixpkgs.xz recompiles all dependents
- staging-next recompiles everything
- downgrading *all* pkgs by 1 month works, but far from ideal
- system.replaceRuntimeDependencies is NixOS-only

Gotta wait for Hydra builds to finish...

account_circle
Danny Lin(@kdrag0n) 's Twitter Profile Photo

this piece by Tina Mai beautifully captures so many of my realizations in this past year

was thinking about it in terms of the butterfly effect too

kernelmag.io/4/chaos

account_circle