noident

If I told my coworker that his or her "brain is a gizmo" and to "not reproduce" I would expect to be frog marched out the door a few minutes later.

Once you have workloads that can't tolerate a power cut + running fsck for a potentially long time, a battery backup becomes an excellent investment. I bought a UPS on eBay for cheap and my home server hasn't gone down since.

No, it's because authentic writing on HN has been drowned out in an ocean of slop, in such quantities that calling it out is becoming an exercise in futility

What can stacked PRs do that a series of well-organized commits in a single branch can't?

If only there were some way to logically break up large pull requests into smaller pieces... Some way of creating a checkpoint with a diff including your changes, and some kind of message explaining the context behind the change... some way to "commit" a change to the record of the repository...

Policies like these are widespread in most companies with >1000 employees

Probably 90% of my coworkers in a US tech company are on a work visa. Now that there is pressure on the H1B program, my company is investing in a permanent engineering team in India. Whether this will pay off for them in the long term is a matter of debate, but it seems like the near-term future of the US engineering team is in serious doubt.

You can't just "dismantle H1B" and expect it not to backfire.

How many of those people have experience with augmented reality? Probably not that many.

$35k seems pretty low for this job. Hindsight is 20/20 of course.

No W. No X. No Y. Just Z.

In fact, the whole article is filled with slopisms, just with the em dashes swapped for regular dashes and some improper spacing around ellipses to make you think a human wrote it.

The LLMisms in the "thinkpad" section caused me to close the tab

Shift-left was a disaster? A large number of my day to day problems at work could be described as failing to shift-left even in the face of overwhelmingly obvious benefits

I clicked the article thinking it was about GitLab. Much of the criticism held true for GitLab anyway, particularly the insanely slow feedback loops these CI/CD systems create.

That's a very unusual and narrow exception involving "foregone conclusion doctrine", an important fact missed by Ars Technica but elaborated on by AP: https://apnews.com/general-news-49da3a1e71f74e1c98012611aedc...

It's often better to say nothing at all rather than to reply with an LLM generated response.

Somewhere along the line the AI bros stopped separating training and testing sets. It's great for impressing the villagers

Ted has some interesting ideas but I personally would not accept any life advice from him

Filippo Valsorda discusses his server for storing age keys

I don't like doing the leetcode grind, but all of the alternatives are strictly worse.

* Take home projects filter out people with busy lives. Wastes 100 people's time to hire 1 person. Can't be sure they didn't cheat. No incentives to stop company from giving you a 10 hour assignment and then not looking at it. The candidate with the most time to waste wins.

* Relying on academic credentials unfairly favors people from privileged backgrounds and doesn't necessarily correlate with skill as an engineer.

* Skipping the tech interview and just talking about the candidate's experience is prone to favoring bullshitters, plus you'll miss smart people who haven't had their lucky break yet.

* Asking "practical" questions tends to eliminate people without familiarity with your problem domain or tech stack.

* We all know how asking riddles and brainteasers worked out.

With leetcode, the curriculum is known up front and I have some assurance that the company has at least has some skin in the game when they schedule an engineer to evaluate me. It also tests your general knowledge and in some part intelligence as opposed to testing that you have some very narrow experience that happens to overlap with the job description.

The author is writing like Java was outlawed or something. There are tons of shitty enterprise Java jobs out there for those who want them. Personally, I worked one of those jobs a decade ago, and the article's description of the "golden age" didn't bring back good memories.

It's easy enough to avoid the NPM circus as well. Just don't put JavaScript on your resume and don't get anywhere near frontend development.

They can and they will. Filing a subpoena for information is a step in that process.

If the WHOIS records are falsified they'll start looking at payment information.

The outgoing Element mobile app (‘classic Element’) will remain available in the app stores until at least the end of 2025, to ensure a smooth transition

https://element.io/blog/mas-migration-unleashes-element-x-on...

I can't find any other communication from Element Creations other than that.

The renaming to Element Classic doesn't bode well considering that Element X still doesn't support a vast number of home servers and a number of Synapse authn/authz features.

If they remove it from the app store, my advice for my users is going to be to switch to fluffychat, and I'll eventually migrate away from Synapse to some flavor of Conduit.

The janky lip sync gave us one of the funniest videos of all time: https://www.youtube.com/watch?v=Js02m-7qHyE

I've lucked out and haven't had any issues like that on my tiny home server.

I wish it were easier to be a Synapse admin! It seems like all server updates are focused on scaling matrix.org and the little guys that provide network diversity are not really being considered a priority.

That's great!

What about Matrix Authentication Service? That's still a separate piece of software that requires PostgreSQL (as opposed to SQLite like my Synapse instance uses) and has no clear migration path for those of us using our own SSO solution (SAML, CAS, maybe your own OIDC provider like Keycloak...), right?

I don't want to run Dex and complicate the stack further, either.

I think 12 hours to migrate is probably optimistic even with built-in sliding sync, actually...

Does this depend on sliding sync and their standalone authz server like Element X does?

I'd love to use Element X, but Element abandoned the form of SSO my community depends on, and I don't really have an appetite to spend 12 hours of my free time standing up sliding sync, a separate auth server, migrating my users to the separate auth server, spending hours explaining to everyone that their credentials live in some other place now, and then migrating my custom server admin software to use OpenID connect. Oh well.

Hi, OP. Any thoughts on why to prefer using your OLTP for queueing as opposed to Kafka? In my mind it would be a drop in ease of observability (though there is still KSQL and a number of UI wrappers) in exchange for much better performance. I'm interested if you had other reasons.

The em dash existed long before LLMs. The fact that 95% of people don't use dashes properly doesn't mean that every single person who uses an em dash is relying on an LLM.

I've run a private server with my friends since the Riot days (2018? Earlier?)

Here's how I made it work:

- No federation. You click an invite link I send you and make an account on my server. Therefore, no spam.

- Encryption disabled in the main group chat room. The encryption experience was very poor in Riot and remains inexplicably baffling to non-technical users in Element. Supposedly Element X fixes these problems, but it doesn't support the type of SSO my server uses yet. I decided to just turn it off. DMs are still encrypted.

- Use Synapse.

- $5 tier digital ocean droplet. I plan to move it to my homelab soon.

For the most part, it has been pretty smooth. Dealing with encryption UX issues was the low point.

I wish that there were custom emoji packs though :(

I had a friend that during his most stressful period as a 24 year old in Investment Banking had strep throat 4 times in 2 months

He may need his tonsils out. I got mine out in my 30s. It was 3 painful weeks to recover but my quality of life is a lot higher.

Who cares if they pay attention to 429s? Your load balancer is giving them the boot, and your expensive backend resources aren't being wasted. They can make requests until the cows come home; they're not getting anything until they slow down.