@RAOF I think there's a bunch of warts in the D-Bus protocol that people would fix if they were to do it over (GVariant is an example, trying to reduce the number of copies needed to decode a message).

There also doesn't seem to be an accepted best practice for doing point-to-point D-Bus: e.g. can you just start sending D-Bus messages on a socket without all the regular connection establishment process?

@bigiain @NewtonMark You wouldn't have had AfterPay willing to give you a loan to make buying that much in one go more affordable.

@bigiain @NewtonMark The first photo in this story is the one that was being shown at the time: https://9now.nine.com.au/a-current-affair/perth-teen-almost-deadly-nang-addiction-wa-premier-mark-mcgowan-response/a5418b85-c328-4bd8-a600-c39eb831a27d

@bigiain @NewtonMark Some of the companies selling these canisters list both volume and weight.

The 3.3 litre canisters seem to contain about 2 kg of nitrous, so I'd assume they're talking compressed volume. The news stories at the time showed many of these large canisters.

So she was doing closer to 0.5 - 1 kg of nitrous a day. As I said, a stupidly large quantity.

@NewtonMark this is a story about the girl who triggered those laws: https://www.watoday.com.au/national/western-australia/this-perth-teenager-tried-nangs-at-leavers-now-she-is-bedridden-in-hospital-20230516-p5d8us.html

That is some pretty serious harm, but she was also using a stupidly large quantity.

@pid_eins I hadn't seen that. Looks like it'd greatly simplify some of the AppArmor policies I've written in the past (switching to a subprofile on exec that can only do NSS/PAM and send the D-Bus method).

@pid_eins I guess that means polkit is never getting absorbed into the systemd project?

(unless there is a good way to replace polkit-agent-helper-1)

@decryption It's second hand, and they're disclosing the possible fault before the sale. So there's probably no basis for a return. With that said, they seem to have the same disclaimer on every one of their eBay listings, so it could just mean they don't test that stuff.

I'm kind of wondering if it is worth taking a risk and getting one as a donation for the local LUG. We're currently stuck with some machines with GPUs that Nvidia doesn't want to support any more, so can't easily be upgraded. I'd want at least a few of those USB ports to be functional though for audio and video capture.

@decryption The "may have broken USB or Ethernet" bit seems a bit concerning.

@patrick the rpm package doesn't include the size of any of its dependencies.

Similarly the flatpak packages don't include the size of the runtime they use. Presumably the larger flatpak packages is using a runtime that includes less of it's dependencies.

@NewtonMark Receiving $1 trillion in value for only $6.5 billion outlay seems like a pretty savvy business decision. What could possibly go wrong?

It's kind of amusing that the local post office has an "Open 24/7" sign next to it's post office boxes when it is located within an arcade that is closed on the weekends and after hours on week days.

@davidgerard On the other side, it's amazing how many features people decided they could live without when shown Google Docs' (then unusual) real time collaboration features.

@NewtonMark I'm also imagining the shit show if an airline had to explain that they had to reduce or remove the overhead luggage bins to make room for people to stand without hitting their heads.

@NewtonMark I wonder if there is a market for keeping the passenger numbers equal, but using these horrible seats to make more room for the sitting seats (and charging proportionally more)?

@mcc For more fun, that at uri scheme spec mentions that a "did:web:..." identifier might be stuffed in the authority section. The spec for those identifiers say they might include percent encoded colons to represent identities served from non-default port https servers.

I wonder how their code handles that currently? Do you need to write "at://did:web:example.com%3A3000:user:alice" or "at://did:web:example.com%253A3000:user:alice"? If the former works, then it might not be possible to fix the encoding.

@decryption It is kind of frustrating thinking back on what was possible ~ 15 years ago w.r.t. self hosting your online presence, and how it's not an accident that we got to where we are.

There are clear usability benefits to the large social networks around authentication, identity, notifications, etc. But we used to have better tools for dealing with RSS and the like.

And a lot of the walled gardens we have today weren't quite as walled off as they are now (e.g. you could give Facebook an RSS feed, and it would automatically make posts on your page from it).

It's particularly annoying seeing the products Google killed to try and drive Google+ engagement.

@benno If it is just a case of "how do I connect to servers that aren't routable locally", would the -J option work? That is:

ssh -J host1 host2

This opens an ssh connection to host1, and uses it to tunnel a second ssh connection to host2. Both connections are initiated locally, so don't rely on agent forwarding.

If you always need to do this for a set of hosts, you can also use the ProxyJump option in ~/.ssh/config, removing the need to explicitly use the -J option.

@mcc This is one area where Go's package naming seems like a win: anything not within the standard library gets namespaced, so no one gets to name squat.

I can see why short package names might be desirable, but I'm not writing out full Go package names anywhere outside of the import blocks at the top of a file.

@redezem @decryption The rollout of new tag-on machines in Perth seems like it is close to complete.

They might not turn on debit card payments as soon as the rollout is complete, but that rollout is a prerequisite: you don't want to let someone tag on with a debit card, only to find the machine at the destination doesn't recognise the card. Or to make one journey on a debit card and not be able to transfer.

I suspect it'll be available next year, or perhaps even this year if everything goes well. It isn't new experimental technology at this point.