dotwaffle @matthew.walster.org · 25/09/2026In the past, on v6, I grew accustomed to just creating a new random address (48-bits of entropy to choose from in a /64!) and then using the port number I wanted, using a listen-addr of that fresh address. Works surprisingly well, even if you occasionally forget to release the address afterwards! 020
dotwaffle @matthew.walster.org · 10/09/2026I know plenty of people with captures of "bad" updates sent that have broken other parsers and state machines over the years if you wanted to put them in the test suite! It's all fairly reasonable until you get to route targets and evpn! 110
dotwaffle @matthew.walster.org · 10/09/2026Especially via ssh over about 30ms rtt, where you can see the cursor constantly fluttering all over the text input to draw the sparkles.. 010
dotwaffle @matthew.walster.org · 05/09/2026I'm assuming `pcrypto` is `pcrypt`? If so, last time I tried it I ended up with less single-stream performance not more ;) It only really made sense as a VPN concentrator. Maybe it's improved in the last 6-7 years though! 010
dotwaffle @matthew.walster.org · 05/09/2026Kernel IPsec is really inefficient though, locked to a single core usually too. 100
dotwaffle @matthew.walster.org · 05/09/2026And where storing 1GB of data for about a year on their service (S3) costs less than the fee they charge for the literal <0.1s it takes to send it from that service to the office down the street or to a competitor. 000
dotwaffle @matthew.walster.org · 06/08/2026I've been meaning to try u-root's `cpu` command for similar ideas... github.com/u-root/cpugithub.comGitHub - u-root/cpu: cpu command in Go, inspired by the Plan 9 cpu commandcpu command in Go, inspired by the Plan 9 cpu command - u-root/cpu 000
dotwaffle @matthew.walster.org · 03/08/2026I'd also like it if the default go mod cache was in $XDG_CACHE_HOME like go-build is, GOBIN was in $HOME/.local/bin (though needs arch awareness...), and that `go clean -cache` had an option to remove anything older than a particular time and/or a size limit because with LLMs those get LARGE :S 000
dotwaffle @matthew.walster.org · 30/06/2026Not just the diff, the release notes too. When there's some new feature, you can ask if it would be useful to the current project, and (with good release notes) you just might find it solves a problem you'd been working around for a while! 130
dotwaffle @matthew.walster.org · 19/05/2026A while ago I decided "no more CI config in repos", never been happier. Though I must admit, I've been meaning to try dagger.io for quite a while now...dagger.ioDagger.ioBuild powerful software environments and containerized operations from modular components and simple functions. Perfect for complex software delivery and AI agents. Built by the creators of Docker. 000
dotwaffle @matthew.walster.org · 19/05/2026The face people make when you say "memory safe language, like Java" :P 110
dotwaffle @matthew.walster.org · 19/05/2026I think I got mine in ~2015, roughly about the same time I started at Fastly. I must dig it out some day, wonder if it still works... 000
dotwaffle @matthew.walster.org · 03/05/2026Highly recommend looking at Typst if you haven't already -- so much easier and so much faster! 010
dotwaffle @matthew.walster.org · 13/04/2026I guess @golang.org may want to know? It's happened multiple times -- both starting around 03:30 UTC on different days strangely enough... 000
dotwaffle @matthew.walster.org · 13/04/2026Anyone else getting really strange runaway memory usage in Go 1.26.2 when using `go doc`? My system ends up swapping and crawls so bad that even TCP connections timeout, including my ssh session... 200
dotwaffle @matthew.walster.org · 19/03/2026Writing a (network) fs in Go/9p. I get sad when I see `pjdfstest`: Tests: 181 failed, 48 skipped, 169 passed, 398 total Error: Some tests have failed Then I try the tests in a container (overlay/ext4): Tests: 39 failed, 48 skipped, 311 passed, 398 total Error: Some tests have failed Happier :P 030