jallmann

> MasterCard and Visa have no business unilaterally, secretly, and unaccountably policing their idiosyncratic idea of moral righteousness

That's not why they do it. The reasons are regulatory compliance and risk. Processors would be in big trouble if they facilitate payments when they shouldn't, or broke due to rampant fraud in certain sectors.

I get that you might not like it, but take it up with the US government. The processors would be happy to move as much money as possible to make as much money as possible.

Charge it up to 80% in the firmware but show 100% in the UI. Wear leveling algorithms probably do something similar already.

I remember buying a Radeon at Fry's (Fountain Valley, Roman Empire), going home, opening it up and... it was something else, a cheap VGA card. They gave me a hard time about taking it back too, 16-year old me didn't know how to cut through the BS so my dad had to go back with me on a 3rd trip and get it sorted out. What a place.

http/1 traffic is a lot easier to inspect

> As patient you need to track who is in an out of network

It's worse than that, there are also "tiers" within a network, even for different physicians under the same health group. We were actually debating moving our kids to a different doc (but in the same office!) because of this.

The whole system is madness.

How does this differ from Pocketbase?

Inflation generally encourages economic activity (spend your money before it depreciates) while deflation discourages it (save your money because it will appreciate)

Lower prices are seen as positive for as long as they stimulate higher overall consumption. If prices fall but consumption does not rise, then most economists see that as a problem.

For those that don't want to run a SMTP server, I built Ephemeral Postal which offers a basic API for polling a mail box and retrieving messages, along with an in-browser UI. You get an entire subdomain to yourself that will take any address that you throw at it. https://ephemeralpostal.com

Not everyone has access to a specialized platform team or technical manager; very often it is a solo (or small group of) devs just trying to Get Stuff Done

A few years ago, I read a slightly updated version of Constance Garnett's Anna Karenina and it was stunning.

I have been reading a copy of P&V Crime and Punishment that I found laying around, and it does not have quite the intensity that I was expecting. Will probably try another translation - deciding on which one to read is half the fun for me anyway.

Sign languages have such enormous variability that I have always thought having fluent sign language recognition / translation probably means we have solved AGI.

Detecting the presence of sign language in a video is an interesting subset of the problem and is important for building out more diverse corpora. I would also try to find more conversational sources of data, since news broadcasts can be clinical as others have mentioned. Good luck.

Many deaf people do prefer sign language as an accessibility option over reading captions, even if the interpreting can be hit-or-miss.

Interesting ... I checked and the difference is closer to 5% ($0.99 May vs $1.04 June), and June's invoice was actually higher than May's. They are billing that amount because the storage amount itself is fluctuating (alternating 66 and 69 GB) with zero writes / deletes to the bucket since around March, and staying well within the free tier for reads. Will probably have to leave it be as one of the mysteries of life :)

I actually have a similar story about Cloudflare, image hosting, billing and support, but the amounts involved were nowhere near as high.

I was evaluating image hosts a while ago. CF Images seemed promising but I was afraid that my use-case would grow to become cost-prohibitive, and I really didn't want to deal with a migration off it later. So I settled on rolling-my-own with R2 and ffmpeg to pre-generate a few renditions [1].

In short, R2 billing is weird. I actually haven't updated my bucket in months and yet the storage amount fluctuates a bit month-to-month. (Is R2 storage accounting probabilistic?!?) I also got hit by a billing line item in May for "R2 late usage" from Feb to March which I could not find any information about.

Their support only recently got back to me, with a very unhelpful / irrelevant answer of course.

The amounts involved are only ~$1 so I don't care too much - I was mostly curious - and I don't blame them for maybe not wanting to expend too much effort looking into irregularities on such a tiny account.

No regrets though. Using CF Images would still be 5x more expensive month-to-month even at my (tiny) scale. The $5-per-100k flat-rate is probably the thing that allows them to make any money on image hosting.

[1] Word to the wise - there are a lot of subtleties around building production-scale image hosting that often times it may be much easier to simply pay the premium for a managed service. For example, what if you want to only generate renditions just-in-time to save on storage? What about updating your rendition list on the fly, eg adding a new one, or even removing older ones from storage? When / how often should you sweep storage of unused renditions?

Also built something very similar - https://ephemeralpostal.com

Comes with a small API so your integration tests can actually check the contents of emails that were expected to be sent out. Did a Show HN a while ago with more details - https://news.ycombinator.com/item?id=40590670

Feels like there is a similarity in a big ball of software mud and complaining about the layout of a website. Does it do the job? Yes? Cool, moving on.

Total layperson here, but here are some things I have been wondering that the article didn't really address.

What happens if something like Starlink is intentionally targeted during an active conflict, maybe with something like a fragmentation warhead? The short deorbit time seems like a feature, if it can deny access to space for the duration of the conflict while minimizing long-term effects. Or is there enough... space... up there that it's a non-issue?

Also wondering about the impact on communications with higher orbits, eg GPS, assuming these satellites even remain intact. If LEO is contaminated, maybe with something like reflective foil strips / confetti, would signals still be able to get through? Or is that again a matter of sheer scale where it is unlikely to cover a large enough area to be a real concern?

This is great news, but simultaneously disappointing that it is necessary. FFmpeg underpins tens of billions of dollars in value for the companies that use it, yet the support from the community is proportionally minuscule.

The specific issues that this article discusses (eg Nagle's algorithm) will be present in most packet-switched transport protocols, especially ones that rely on acknowledgements for reliability. The QUIC RFC mentions this: https://datatracker.ietf.org/doc/html/rfc9000#section-13

Packet overhead, ack frequency, etc are the tip of the iceberg though. QUIC addresses some of the biggest issues with TCP such as head-of-line blocking but still shares the more finicky issues, such as different flow and congestion control algorithms interacting poorly.

These types of keepalives are usually best handled at the application protocol layer where you can design in more knobs and respond in different ways. Otherwise you may see unexpected interactions between different keepalive mechanisms in different parts of the protocol stack.

> Most people would see this as wasteful NIH but I'm convinced it was a net benefit for them—they managed to be so productive in an absolute sense and especially on a per-engineer basis because they were willing to build so much themselves, not despite it.

It paid off for them, but I would say it's a risky move in general - it is very easy to get sidetracked with failed projects that have nothing to do with the business. There are many examples - Uber's internal chat system, etc.

Others tools were more a necessity, eg the OCaml build story was not great back then. Investing in OCaml itself was a calculated risk.

Also I wonder how things would be different if they were starting today, where there are more mature tools in the space.

Curious - was this for Microsoft internal services or customer deployments? If the latter, how did they know what was PII?

I use nixpkgs on MacOS, is nix-darwin is a different project?

I love Nix but it probably has too many rough edges for the typical homebrew user.

My Twitter feed has a bunch of people doing extremely niche deep dives into Cold War weapons systems. For example, on the design of MiG-23 air intakes: https://twitter.com/BaA43A3aHY/status/1753715489686057384

What an excellent write-up. Very clearly explained, and the use of gifs and visuals to illustrate concepts was spot-on. I'm only a software engineer with a limited understanding of this field, but really enjoyed reading this and learned a lot. Well done and congrats on the PhD.

Congrats on the launch - rooting for y'all!

Multi-factor authentication for your Github PRs. https://otpguard.com

BoringAvatars [1][2] or Robohash [3]

[1] https://boringavatars.com

[2] URL service: https://github.com/boringdesigners/boring-avatars-service/bl...

[3] https://robohash.org

Curl and most libraries don't unwrap SSE for you out of the box, you lose the one-shot request-response leading to nonstandard flow control in integrations, etc.

Sure you can find random libraries to help but they do need to be specifically built for SSE. Plain (long-)polling does not need any of that, which makes it more attractive from an API interoperability standpoint.

Sorry, that part of the comment was probably not clear - I was comparing "plain old polling" (stateless request-reply with no delay) with "push", ie long polling