one account wherever you go ☺️
your handle isn’t just for one app anymore !
we took @hello.pckt.cafe for a walk across the atmosphere
07.03.2026 05:37
👍 337
🔁 77
💬 16
📌 25
one account wherever you go ☺️
your handle isn’t just for one app anymore !
we took @hello.pckt.cafe for a walk across the atmosphere
I tell people we offer three neins reliability, in German.
Is the service reliable? Nein, nein, nein.
It has been so exciting to see @npmx.dev come together over the past month. We're proud to offer our support — this project is just getting started. Congrats on the launch!
atproto.com/blog/npmx-al...
My feed is a never ending @npmx.dev celebration, as it should be! Congrats to the release!
Congrats!
Just a short piece of how I use @obsidian.md daily notes, how it helps me with standups at work and staying on top of things in general.
”The distinction between programmer and user is reinforced and maintained by a tech industry that benefits from a population rendered computationally passive. If we accept and adopt the role of less agency, we then make it harder for ourselves to come into more agency.”
Back from vacation and feeling properly energized, time to be more active here again as well :)
I wrote a deep dive on CIDs, the self-describing cryptographic fingerprints behind every record, blob, commit, and Merkle tree node in ATProtocol. Part 1 covers the history, the encoding primitives, and the specific constraints ATProtocol applies.
screenshot of PDSI showing npmx.dev moved to our npmx.social PDS hosted in Europe Service updated #atproto_pds − https://pholiota.us-west.host.bsky.network + https://npmx.social
hello from our new european home 👋
I just learned about @bookhive.buzz imported my goodreads library over with 118/121 titles being imported successfully
bookhive.buzz/profile/mari...
tfw you install helm on nixos and instead of deploying stuff on Kubernetes you suddely have a synthesizer
BlueSky, atproto and the people building here really give me hope that we could make the internet fun again
The Good Internet – Director’s Cut How to escape the algorithmic doom loops and reclaim our digital lives www.sachajudd.com/the-good-int...
Very much yes
atproto is cool because it is fundamentally pragmatic about what it means to be decentralised. decentralised where it matters, simple and scalable where it doesn’t
And now I've got my whole nixos configuration as a flake up on tangled!
tangled.org/mariuskimmin...
And thanks to this video by vimjoyer I even got gaming with an nvidia card to work relatively seamless
www.youtube.com/watch?v=qlfm...
I installed NixOS for the ~5th time on my personal computer again after using it on servers for a while and this time I may just stick to it. I went the no-desktop path on the installer, added niri and noctalia-shell and it's been sooooo smooth
This is the most succinct version of a post I've been trying to write for two months. When I stop for just a sec to think about the possibilities of The Atmospheric web, I'm a little overwhelmed by how exciting it is. Come build it with us!
> To avoid such occurrences in future I would like to recommend you to change the value of "M out of N" for your alarm to something like "3 out of 3" as its currently set to "5 out of 5"
Or, you know, just actually alarm us at 5 out of 5?
Rant over.
Like what do you mean "CloudWatch looks for additonal datapoints"? Where? Why?
And how are people fine with having non-deterministic alerting. I might as well let an LLM decide if this metric is Alarm state. Or toss a coin.
> For example, if the alarm is having datapoints to alarm as 5 out of 5, it should look for 5 data points. But in the backend, CloudWatch looks for additional datapoints as well. In this case, suppose the alarm looks for additional 3 data points.
> However, CloudWatch Alarm takes additional data points as well into consideration to avoid false alarms. Due to which, even though your metrics was not having datapoints for more than 5 minutes, the alarm still did not transition to "In Alarm" state.
> I would like to inform you that CloudWatch Alarm evaluates every minute irrespective of the time period you select, and the evaluation is based on the window of time defined by the Period and Evaluation Periods. In this case, you have kept the datapoints to alarm as 5 out of 5.
Now, we faced Cloudwatch shenanigans before where it "corrected" past data as new datapoints come in but this case was even more strange. So we decided to open a ticket with the support:
I have gotta rant about CloudWatch (AWS) for a moment - we have an alarm setup for 5 out of 5 datapoints where missing datapoints are treated as breaching. Now over the last days a case happened where this should have clearly triggered - we are missing 10 datapoints here. No alarm fired.
www.seangoedecke.com/how-i-estima...
This feels true, estimating how long a task will take is the most useless thing we (as an industry) keep doing. It's pure theater and also so draining.
2026 is the year Bluesky and the Atmosphere really come alive
here's what's next
bsky.social/about/blog/0...
Don't you love it when managers declare multiple things as "the highest priority" and then wonder why not all of them get "enough attention" if tho they set the prio so high?
