Notices by Hailey (hailey@hails.org), page 2

@whitequark it is in fact a skill issue if someone yanking their random little package causes problems for you and we should rub it in lmao

enjoyed reading this piece on perl and its decline https://www.beatworm.co.uk/blog/computers/perls-decline-was-cultural-not-technical

hypothesizing that having exposure to Perl or not is a kind of programmer generational cleavage

@whitequark I was just thinking actually, whats your thoughts on a feature to reverse proxy on 404?

@whitequark oh that's quite cool! does the patch persist until the next git push?

>he tells me his language has an exciting new design for asynchronous io that solves the function colouring problem
>ask if it’s just threads or something genuinely novel
>he doesn’t understand
>he says it’s a good asynchronous runtime miss
>read his blog post
>it’s threads

imo function colouring as a problem is akin to calling it road colouring when there’s a separate bike lane you can’t drive your car on. It’s not a problem to be solved, it’s the whole point!

Back in the early days of GitHub, git over ssh was terminated by a hacked OpenSSH which looked up public keys for the git user in MySQL rather than from an authorized_keys file. That OpenSSH was the system sshd on the same frontend web servers which ran the Rails app, so if you were staff you could get a shell on a random frontend by typing `ssh github.com` and you were straight in

I am quite an experienced Linux user and would like to contribute to Debian the same way I regularly contribute to all sorts of projects as I go along scratching my own itch, but I just cannot get my head around its development

the massive issue I have with debian, despite it being my preferred server OS, is I have no idea how you’re actually meant to do anything with the platform beyond consuming it. I build my own packages for my infrastructure because it’s more robust than copying tarballs around, but it’s complicated enough that I have to rely on a third party tool (makedeb) to do it. And here an APT maintainer is claiming that apt-ftparchive sees no serious use outside of Canonical. How are you supposed to host a package repo then?? Inscrutable system https://lwn.net/SubscriberLink/1046841/5bbf1fc049a18947/

@fraggle I wonder how much of the fragile mess we have downstream of the OS (containers, curl | sh install scripts, distributing statically linked binaries, language specific version managers) can be traced back to “it’s really hard to build packages and we’re just trying to get shit done here”

do you often pipe curl to jq and have trouble when you also want to see the response headers with -i?

hack of the day: curl -s -D /dev/stdout -o >(jq) https://url...

`-D /dev/stdout` dumps headers to stdout
`-o >(jq)` uses bash process substitution to pipe the response body separately to a jq process

@whitequark you get it

Weighing in on the cloudflare unwrapgate against my better judgement. I wouldn’t have written unwrap there. But it’s an error loading config on app boot so odds are that even returning the error “like you should” is just exiting with a nonzero status with extra steps. You could recover by failing open but that could have easily masked the problem, not great for a WAF. So basically my take is unwrap is a red herring. The real issue is the bad config deploy and the seeming lack of a canary deploy or rollback to known good config. Standard ops stuff basically 🤷♀️

Remember when people thought writing tests in cucumber was a good idea

@whitequark no idea if it still works like this or whether it's all on azure object storage or whatever now, but i was pleased with how neat the solution ended up being once i had the brainwave to use gc, and i don't recall it ever having any major problems in production once shipped

@whitequark no particular comments on the gc algo, you're using gc in a a very different way. we didn't attempt to do any object dedup for pages, whole site builds were just rsynced from build server to file servers. mysql was the source of truth for which file servers held site replicas, new builds were allocated to the N file servers with most free disk at time of build, and each file server ran GC to clean up dangling replicas at some point in time after the replica records were deleted from mysql. idea was we'd need some kind of gc anyway because we can't delete from mysql and the fileservers in one transaction, so skip synchronous fs deletions entirely and make gc load bearing thus continuously proving it works

@whitequark fun to see you use garbage collection for grebedoc. I also landed on garbage collection as the right solution when reworking gh pages to do its own replication rather than drbd back in the day. very different set of design constraints and different kind of gc, but interesting to see it recur as the right solution for this kind of thing

@whitequark ++ in general for any tool and any art. I have a windows pc and a basic ass microsoft keyboard and mouse at work and I’m getting shit done

it’s a small nugget of the whole thing but it’s so funny to me that David HH has written not one but multiple posts about things that are better than going to therapy. one of the crash outs of all time