You might also want to check the latest Ladybird update: https://youtu.be/cbw0KrMGHvc

Doing a rough comparison: 500k KWh * 365 / 240 billion KWh= ChatGPT uses ~0.076% of the power that Crypto uses.

Just to put Crypto into perspective.

PS: The 500k is from the article (which states it’s per day), and the 240 billion for crypto is the first result I saw from Google, so take with a grain of salt.

edit: I only just now saw which space (?) this is from. Please don’t bully me, I’m only here for the discussion (which so far has been civil).

I just sometimes wish there was a little more content getting posted.

Well, that’s a problem that’s within your power to fix! Start posting!

Nonprofits can “own” for-profits.

One of the saner reasons for this structure is that the non-profit owns the things the for-profit works on. If the for-profit goes under, all things are still owned by the non-profit, so some large tech company can’t swoop in and yoink anything available.

This includes any and all data generated by the for-profit, which means your data is “safe”.

Oh wow, looks like the Haskell devs have been hauling ass! Nice!

I remember the language server being a thing already, but it was in some alpha stage back then. Good to know it’s usable now! :D

Here’s what I remember from Haskell (around 2018):

I love the language, but hate the tooling.

Used it for Uni (did a minor where I learned Haskell, recursion, parsing and regex - probably the most information dense part of school I’ve ever had. Half a year of minor also burned me out, so I never went for my masters; I’m OK with my Bachelors :D ), but never felt like picking it back up.

Multiply by Phi (1.6) / the Golden Ratio (still 1.6) to get km.

Typically, this is what you need to know:

100 miles = 160 km 60 miles = 100km

I just learned the Faroe Islands is connecting all their islands (Exploring New Road Tunnels and Power Generation in Faroese Islands). Even to islands that only have 500 people living there.

I don’t see why the USA couldn’t increase its railway (hell, you can learn from Japan if it needs to be earthquake resistant) starting today.

lol you beat me to it: https://i.imgur.com/0h892RM.png

I haven’t used 8GB since… 2008 or so? TBF, I’m a power user (as are most people on any Lemmy instance, I presume), but still…

And sure, Mac OS presumably uses less RAM than Windows, but all the applications don’t.

Interesting read, but does it beat JPEGMini? Yes, it’s a paid product (not that I’ve paid, ☠🏴 yarr), but it does what it needs to do, and it does it well.

Oh, oh, oh! I got one not mentioned yet:

ELK.

Well, not the whole of the ELK stack (Elastic, Logstash and Kibana, though the full stack size is much larger nowadays), but their watchers. A watcher is a piece of JSON with some search specifications on when to trigger and send an alert to email/slack/teams/whatever. We’re basically abusing it as an alerting system, and generally it works… Fine… Presuming Filebeat actually ingests our logs (which is partially our fault, as there’s a fix, but it takes too damn long to drag 3 teams along to implement what needs implementing to fix that problem).

Anyway, the problem is not the watcher itself, even though it is painful (heh) to learn the structure. It’s “Painless”, the JVM-based scripting language available in a watcher. It’s anything but. It is SO painful to write code, inside of a JSON object, making sure everything is exactly as it should be, having to use the DevTools in Kibana to try and trigger it, wait to see what enormous error comes out while praying it works. No IDE, no nothing. Ah, I lied. It does have Syntax Highlighting, for non-Painless code, IIRC…

Oh, having to dig information out of the data you get is super unintuitive too.

At least the UI/Kibana is good, and Elastic is pretty good too. Fuck Filebeat though. And Painless.

The background noise surpression of Teams is peak quality (vs Webex and Slack, though Webex is somewhat good)…

Did you check the calender in Teams? Not to be confused with the calender in Outlook, which may or may not overlap.

I’m pretty sure you can disable the bleeps and bloops people are complaining about - get a tool, then don’t spend any time to learn it and then suffer using it.

Look at mr fancy, with his interface that can search through topics - we’re just flying by the top of our head >_>

At least we were smart enough to just generate a CloudEvent and jam it into a KafkaMessage before sending it, so that’s pretty standard.

We just use it as a messaging bus - we’re not sending “big data” through Kafka. We tried, failed, and immediately gave up, lmao. Now we just save data on HDFS and send where we stored said file (effectively a “claim-check pattern”, if you’re into design patterns)

rockstar

We fixed that one: https://codewithrockstar.com/

I ran the PowerDelete tool last night to nuke my profile

Note that the profile only shows the last 1k comments - I did a GDPR request to get my data (took em a few weeks - they have a month the time to do so) and then used that data to find all my comments and replace a whole bunch of comments with GPT-generated bullshit about spez being a poopyhead.

I won’t pick up my shit and move, I’ll kill the plants, salt the earth and poison the well. Fuck 'em. It’s all for shittificating the site for YEARS.

I still charge to 100, but I use a slow charger, so my phone doesn’t start to spew flames while it’s charging. I wouldn’t be surprised if that helped as well (as heat is another battery killer).

I just can’t be bothered to handle that shit manually.

Here’s a decent impression of the times: http://i.imgur.com/mAUyo.jpg

But back in the day (2003-ish) we still had amazing things to look forward to:

• translucency (windows were not see-through)
• realtime lighting and shadows (shadows were blobs below a model)
• metallic reflection, and reflections in general (though working mirrors existed since at least Duke Nukem 3D, but those were a hack; copy the room and player model and flip them around to create the effect of a mirror)
• further viewing distances (though this isn’t a positive, IMO)
• physics (everything was static - models moved, but did not rotate (much))
• inverse kinematics

It’s crazy how far we’ve gotten, but view distances spoil everything (IMO), and graphical improvements have slowed down (not stalled, but definitely slowed down) with Ray Tracing becoming wide-spread being the last big graphical improvement (since 2018).