FWIW, every Node.js distro I’ve ever used leveraged OpenSSL for those, under the hood. I imagine other runtimes do too.

I hope so. Crypto is a weirder beast than some things because numbers can look random and not be without much, much deeper inspection: https://www.openbsd.org/papers/bsdcan14-libressl/mgp00001.html

https://nostr.com/

No idea what the Google login thing was. If I had to take a wild guess, maybe you found some place that also offers some key management/signing features as a way to make things more user friendly?

I’ll buy that. I just have trouble squaring things like that with things like gpg-agent and how they cache passphrases and keep them from swapping to disk, etc. I’d want to know more about it and whether it’s generated in the browser or on the server. I’m also not even sure how I’d go about doing some sort of review of the javascript algorithms or where their random numbers come from. I’m not throwing shade on them; I just don’t know.

iris.to […] Primal.net

I’ll look.

Hard pass for me I think on nostr. The signup says to continue with google (heard of 'em) or NIP-07 (not heard of 'em), and the generate nsec button offers me an opportunity to download “my” private key. The file has a keypair and a helpful message that I can share the public key, and that I should never share the private key that it just sent me.

Maybe there’s something I’m missing, but all the previous systems I’m used to use that to mean that you generate the keypair yourself, and then you can post the public key publicly while you should always keep the private key private which breaks the guarantee of the math that only the person with the private key can undo the message that the public key locked.

I feel like there’s something I’m not understanding.

Correct, except I signed my public key. You now have cryptographic proof that I’m me. And as me, you can take my word for it that I only made that key to see what signed messages would look like on lemmy so that I could post in these sorts of situations and so that I could test Kleopatra. It’s not intended to be used for anything serious.

It should also enable anyone who wanted to send an encrypted message to me to do so.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

We could all do it today.  The technology has existed for a
very long time now.  We somehow allowed the market to get to
a point where the crypto mostly benefits the corporations
using people as a product.

- -----BEGIN PGP PUBLIC KEY BLOCK-----
Comment: User ID:	lemmy_posting_key
Comment: Valid from:	7/31/26 2:00 AM
Comment: Valid until:	7/31/29 12:00 PM
Comment: Type:	255-bit EdDSA (secret key available)
Comment: Usage:	Signing, Encryption, Certifying User IDs
Comment: Fingerprint:	3965EF2558DE27BDF057EBD6D1800E8356A14454

mDMEamxIKxYJKwYBBAHaRw8BAQdAHa59Q90fLXONJ/dheGvaTUsh8RxpYbN5/kLk
VwhuJ8S0EWxlbW15X3Bvc3Rpbmdfa2V5iJYEExYKAD4WIQQ5Ze8lWN4nvfBX69bR
gA6DVqFEVAUCamxIKwIbAwUJBaV4ZQULCQgHAgYVCgkICwIEFgIDAQIeAQIXgAAK
CRDRgA6DVqFEVPHlAP9zcGI0AvP/I1fqIYfks8gArPWsbbXFMA3xdXuxpB0fkQD+
NEKRbjZaoyQ5MU2IAlHdfC0DWCL7gyJ/aXtJQZPOAg24OARqbEgrEgorBgEEAZdV
AQUBAQdA0itDkHmO4JKPtyK0+c+7KvaB1ub/ees6koIZo2S13wEDAQgHiH4EGBYK
ACYWIQQ5Ze8lWN4nvfBX69bRgA6DVqFEVAUCamxIKwIbDAUJBaV4ZQAKCRDRgA6D
VqFEVIaOAP9FUfaxNBI+gsBA4s6INSqhWCxhRPWCgQAk0wltsgpnkgD/bfpilGAx
XBvfnJBlHbel5z73T4IEBqIvDTHX6t9AUQo=
=kclm
- -----END PGP PUBLIC KEY BLOCK-----
-----BEGIN PGP SIGNATURE-----

iHUEARYKAB0WIQRUzy5RsYePpEjLPsQmKLxeU5C8sgUCaqIEbwAKCRAmKLxeU5C8
soZBAQCuGIZLW0SKpqqNrVOheG8U/t8YLrvLSal/MI5f3HgfowEAr6AV0lsrLmlV
5BIldH/5q5WIbiZOiyPM4O/oerODSQQ=
=jnr7
-----END PGP SIGNATURE-----

Websites have to search for new revenue streams. They don’t just walk into a corner of the internet and think “drat, someone already has a forum for that!” and walk away. They come up with some unique twist people won’t have normally thought of and run with that until it stops working. See this guy’s comment below:

• https://lemmy.world/post/50436416/25191964

With reddit they need to be able to show that their advertisers aren’t just serving ads to bots (although we’re seeing ads for bots now apparently). They have to prove a certain number of real human beings saw something thus the login requirement.

Now, people used old reddit because new reddit sucks. New reddit just shows you the top few posts, and makes you expand replies, thus proving you’re something they can bill for. This also serves the purpose of pushing nuanced, long running discussion down to where people don’t see it. Just look at the cat picture, upvote it, and move on.

And for the record reddit has been crap for a very long time. At least since whoever that lame ass admin was who went into the backend database and edited someone’s raw post, and if not then, then Victoria getting the axe should’ve been a red flag to everyone that you weren’t a member of reddit…not really. I know there were some local meetups in that timeframe. How awkward do you think the AMA local BBQ was? Do you think reddit cares?

I did a quick google search, and there’s a C++ boost library and a Windows API if you needed to do that. Might have a harder time finding english documentation about it, though.

Sure.

For a while now, if you needed the time from your computer, and it was running UNIX or a variant, you’d use something like gettimeofday() (it’s clock_gettime() now) which would hand you seconds since the epoch at midnight January 1, 1970 in UTC and microseconds/nanoseconds depending. You’d take the seconds, wash it through localtime_r to adjust for the local time zone, and you’d have a struct you could feed to strftime() for formatting.

So my beef with time formats, and what I presume ChickenLadyLovesLife’s beef is, is that when the time format is “wrong” for whatever reason, given the similiarity between the backends of modern time systems, it’s “wrong” for reasons other than technical limitations. You’ve got a count in seconds since the beginning of time, and a function that’s been around for decades that can do whatever stupid stuff you want. Military time, month printed out, day of week printed out, two character years, four character years, seconds, julian day, Doing something silly like day of year, hour of day, and am/pm (“%j %I %p”) is just as easy as something practical like a large fluffy timestamp, or a minimalist four characters military time (“%H%M”). The hard part was getting the time from the GPS or cell phone tower or wherever. The easy part is the time format. Windows also exists, and uses something similar. Modern programming with C++ or Python will have libraries (<chrono> and datetime, respectively) that work the same way, and both those examples use strftime() semantics.

So, if you’re unhappy with the way your device prints out time, the reason behind it isn’t the clock part of it, it’s the display part. There was some rationale on Android’s behalf for adopting the hour over minutes style, but it wasn’t because their available horizontal character limit dropped from four (five if you count a colon) to two. Maybe they can get more data widgets displayed around the time in a little halo? I don’t know, I don’t use android stuff anywhere. If it was apple I’d assume they did UX testing about half height characters and readability.

So, now to the “end goal” part of what I wrote before. The hard part is done and a solved problem, and most time formats work fine for what they do. tmux is giving me (“%H:%M %d-%b-%y”) which works fine, and my desktop environment (DE) is using (“%a, %b %d %H:%M”) which is fine. The other DE is using (“%b %d %H:%M”) which is fine. All perfectly usable, all used the three character representation of the month avoiding problems between dd/mm/yy or mm/dd/yy, and all of those are adjustable to an extent with two of the three kicking back to strftime() if I needed it. So what did android get when they paid developers to make the change to hours on top of minutes? Is there a study I missed that of a group of people who’d never seen a clock before and had no concept of time found the “%H\n%M” style more easy to read? If that’s true, then was it worth irritating the subset of people who are going to care about the change to pick up the other group? I personally don’t see a point in paying to make the change unless they’re expecting to make more money from it, which at the core of a lot of things in the world these days. Here’s something that worked fine, but now it’s being changed to work differently but no better.

Try needing Julian day. If you’re coding then strftime() has supported it since 1989. All you need is “%j” and there it is, with leap year handling, etc.

The end goal isn’t a good/flexible/powerful user experience for the person who bought the device; it’s a larger user base that can be marketed to The Machine for more money out the back end.

RHEL9 and forward require v3, and the numpy in pip as of a few versions back uses either v2 or v3 instructions, so v1 is silently broke for certain workloads. FreeBSD works on it just fine as do Debian based distributions as long as you don’t need recent versions of numpy, but there’s no telling what else out there just tries to run and fails with an illegal instruction.