oh, Slack runs on AWS? I guess that would explain why colleagues felt so much slower, and why it went away for me earlier. Of course it does. Suno runs on AWS too. One day there's going to be an outage of these, and the cloud engineers will tell us "oh, we've restored the outage! We've done what we could!" and,... Nothing. Then 2 hours passes. "oh, We restored it again! It's coming back, we swear!" And yet, the deafening silence of the internet and tanking of markets will be their only proof that no, in fact, nothing had restored itself. #TruthToBeSoon
This entry was edited (2 weeks ago)

Trump’s DHS appointees have dismantled civil rights guardrails, protected agents’ anonymity and encouraged them to wear masks, threatened groups that stood in their way and overwhelmed challenges to their arrests.

propublica.org/article/trump-d…

#News #USPolitics #Government #DHS #Immigration #ICE #Trump

us-east-1 going kaput: the bit that always goes kaput in a globally affecting way is a bit that allows distributed networks to function. i think it's very important to talk about how we can improve dns or bgp or whatever particular aspect goes kaput on a particular day, but it's weird to consider distribution itself as the solution. this is a distributed network! it broke in this way because it's distributed and the bit that allows distributed things to talk to each other broke.
in reply to Matt Campbell

@matt right but that's the crux of it, you're talking about 'central control' but it's not as simple as distributing some computers. it concerns things like "do all these machines run the same software", "do all the isps use the same router hardware", "can barry bezos push a change simultaneously to all these machines". distribution only solves things like "can a single fire / power outage take out half the internet". and then there's things like undersea cables.

I'm quite proud of the fact that I usually find out about downtime of #bigtech services from the news.

For years, almost all of my personal online tools have been self-hosted or run in data centres of small regional providers. They aren't immune to downtime, but when they're down, it isn't half the internet.

Monocultures may seem efficient, but they aren't resilient.

#AWS #AWSOutage #Amazon

Peter Vágner reshared this.

Microsoft Word is the best word processor out there! As long as you're not a screen reader user. If you are, you can't use the web version, and the solution to that is to install the desktop version and then fuck around with advanced screen reader settings until it works somewhat OK. Aren't large trillion-dollar companies that have enough money to change the world for the better but instead throw it into inshitifying amazing?

Do I know anyone (or anyone who knows anyone, pls boost) who has had any joy enforcement of the Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018?

The Equality Advisory Advice Service (EASS) who formally accept breach reports seem to be a waste of oxygen. I'm litigating separately via the Equality Act 2010, but I shouldn't have to.

The issue is around 'perceivable' and terrible fonts + WCAG colour contrast which is not being fixed.

In non-infosec news, I'm curious about Mastodon's opinion on the preferred path forward in mobile device development.

Which would you prefer as the path forward to escape the Apple/Google duopoly?

  • A fully open-source Linux stack with native apps (0%, 0 votes)
  • A fully open-source Linux stack with PWA web apps (0%, 0 votes)
  • A hybrid Android stack, with open native/web apps (0%, 0 votes)
  • iOS/Android is just fine (100%, 1 vote)
1 voter. Poll end: 2 weeks ago

This entry was edited (2 weeks ago)
in reply to Mikołaj Hołysz

@miki Interesting. Seems like this is a risk for any ecosystem that doesn't have a walled-garden app store though right? This isn't unique to PWAs.

So perhaps the next question is, what is the best way to independently verify/vet/choose well-behaved apps?

Or put another way, is it possible to prevent bad-actor apps (PWA or native) without requiring a walled-garden app store?

Seems like this is a requirement before moving forward with anything.

In which I suggest an option for more #accessible command line output to an #openSource library maintainer, and they plus a community member implement it because they're gifts to the world.

github.com/BrianPugh/cyclopts/…

enjoying this part 2 interview with Daniel Kahn on Radiant Others featuring some great deep discussions about the beauty and challenges of performing unashamedly leftist #Yiddish music and finding meaning in old songs. and some old gossip I didn't know about 😅Like I didn't know a disgruntled audience member literally tried pulling the plug on the debut of Dumai at a #klezmer festival😳
radiant-others-a.blubrry.net/d…

#DanielKahn #Songwriters

End of 10 na OpenAltu: příležitost přejít z Windows 10 na Linux

Konec podpory Windows 10 přináší zajímavou příležitost pro Linux, protože nechává bez podpory miliony počítačů, které by mohly dál sloužit a na kterých by mohl běžet právě Linux. Proč přechod na Linux dává větší smysl než kdykoliv před tím a jak jsme se rozhodli tomu jít naproti na OpenAltu.

#desktop #Endof10 #Linux #OpenAlt #OpenAlt2025 #Windows

blog.eischmann.cz/2025/10/20/e…
(reakce na tento příspěvek se může zobrazit jako komentář pod článkem)

This entry was edited (2 weeks ago)

In the wake of the latest AWS US East 1 outage (of many over the years), I think it's worth asking how big is too big for a single data center. Are the DigitalOcean data centers too big? What about OVH? Going all the way to self-hosting on a small server in one's own home is clearly not the right answer for most people, and certainly not for commercial services. But how far should we swing toward decentralization?
in reply to Matt Campbell

Maybe the ideal balance between decentralization and efficiency (particularly energy efficiency) would be lots of little data centers where each one is a single full @oxidecomputer rack. A single fully-loaded Oxide rack has a _lot_ of compute and storage, and is optimized to be meaningfully more energy-efficient than the equivalent number of 1U or 2U servers.

If there were a local-ish VPS hosting provider running on an Oxide rack, I'd rent from that provider.

in reply to Matt Campbell

The Cubbit project started doing something like this for data storage. Nowadays they do provide an S3 service where all data is distributed redundantly across the world. When the project started it required the community to buy one of their tiny devices that you plug into a LAN port and power. That one contains a HDD and software to become a pod for the service. With that you also get some storage space that you can use for yourself. Nowadays you can rent them for storage too.
This entry was edited (2 weeks ago)

sorry internet, i forgot to give you something to go into monday with yesterday.

and as a result, us-east-1 had dns issues and of course, everything went down. because the promise of cloud was multiple zones that create redundancy and prevent exactly this sort of thing from happening right?

its like gloating about having a car with multiple side airbags, but then pulling the fuse out of them so they never work

AWS: We're having an outage that affects only one of the regions in our globally distributed infrastructure platform.
The internet: it's a good thing you don't have, like, one region you consider globally available, that acts as a centralised management point for a lot of your internal workings, right?
AWS: ha ha right yes that would be bad sweatdrop sweatdrop.
This entry was edited (2 weeks ago)

Interesting article about how LLMs actually work: harysdalvi.com/blog/llms-dont-…

One thing I take away from this is that while "grand theft autocomplete" is a very catchy meme, the last word oversimplifies what the LLM is actually doing.

Mike Gorse reshared this.

Just watched this great presentation from @AdrianVovk at the @allsystemsgo conference:

youtube.com/watch?v=uCAlzx_x6r…

It covers the migration to Wayland-only, and of the @gnome session to systemd, eliminating tens of thousands of lines of ancient C code (often a quarter to a half of the codebase depending on the module) and providing more featureful session handling, including session saving and processes suspension. The remote desktop implications are interesting too.

#GNOME #systemd #Wayland

From a different party:

> The party would prioritize the extension of the Metro's Orange line to the Bois-Franc REM station

I hate to brake it to you buddy, but it's when you were in charge of the STM that you should have gone to bat for this. The TBM has been gone for a while. And you know what cost for that kind of project? The TBM.

(context: they built the underground garage for trains past Cote Vertu, going basically half way the distance to connect to Bois-Franc, with a TBM)

#polMTL

This entry was edited (2 weeks ago)

President Trump’s White House is already the second most-blocked account on Bluesky, just days after joining the platform. The only account ahead of it on the block list? It belongs to VP Vance. mediaite.com/media/news/the-wh…
in reply to Hubert Figuière

> It would work with Montreal police to train Metro security officers in "frontline intervention tactics."

> The party would create a police squad targeting violence and drug trafficking downtown. It would invest in video surveillance, [...]

newsinteractives.cbc.ca/featur…

#polMTL

in reply to Hubert Figuière

Also

> The party would replace single turnstile entrances with full-height turnstiles.

They must never have tried these. 1. they are inconvenient 2. cost money that could be used to other things 3. make it even worse when fare evasion happens. Instead of some jumping, they assault by sliding being someone with a fare....

#polMTL

This entry was edited (2 weeks ago)

Good morning fediverse! Been a wile--rough past few months including 2 deaths in the family, a cross-country move and house consolidation with my partner and our cat, etc. but I'm still here. Trying to pop my head back up, catch up on a bunch of replies I owe folks, start socializing online again, and more.

I live in Michigan now. We have autumn for the first time in my life! (Or at least, for the first time in my adult life.) I'm unusually excited about that. And geese are apparently evil incarnate which surprised me--I still kinda like them. Living up north is weird in a lot of ways.

Tomorrow I take an intro to leatherworking class. That and finally relearning how to play my keyboard competently are my 2 dark-winter projects.

Hello Players. Now that our maintenance is done, a long awaited change i was hoping to make. The scrabble dictionary has been expanded. There is a ton more words for you to build, and hopefully a lot of the requests we have for words people couldn't find are now implemented. The game is now available again. I hope this brings a lot more enjoyment and as always if you have any suggestions, please don't hesitate to submit them with the submit a suggestion command via the F1 menu on the client.

Good morning players! Just a brief update after yesterdays maintenance. The RSGames web client is now back up, as well as RSVoicechat. Both services can now be accessed in the usual way you would normally do so. We are continuing work on scrabble and you should see this later on this afternoon. This completes server maintenance officially. If you are still experiencing any issues, please don't hesitate to reach out and let us know.

👁️ Know your Standards
Updated 20/10/2025

"Regularly people link to or quote from some outdated version of a specification. This can be confusing and detrimental to understanding, if the information is no longer correct or relevant."

#a11y #HTML #ARIA #webStandards

html5accessibility.com/stuff/2…