@jamie@zomglol.wtf
@jamie@zomglol.wtf avatar

jamie

@jamie@zomglol.wtf

I build distributed systems and try to make life just a little bit more bearable for the people around me.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

tixie, to random
@tixie@guerilla.studio avatar

iTerm, can you NOT PUT FUCKING AI in my terminal. I don't need AI in my fucking terminal ffs.

Damn, tech bubble is fucking rotting everything

jamie,
@jamie@zomglol.wtf avatar

@konnorrogers @tixie I literally opened Mastodon specifically to post this screenshot 🙃

nyquildotorg, to random
@nyquildotorg@fedia.social avatar

"[LLMs] owe a lot to non-commercial user content: Reddit, Stack Exchange, Quora, Wikipedia, and the "small web".

Ok, but Reddit, Stack Exchange and Quora are not non-commercial 🤷

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg The sites themselves are definitely commercial, but is that true of users' posts, too? Like, are users making money off their posts like they do on Substack or YouTube?

Honest question, I assume not but I recognize that I don't use them enough to know that for sure.

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg I see what you mean now, that the user content is commercialized because the sites make money off of it, not because the users do.

ramsey, (edited ) to random
@ramsey@phpc.social avatar

I managed to avoid for 10 years, but it’s finally caught up to me, so I hope I’m a Kubernetes god after going through all this required (by job) Kubernetes training.

When do I start honking like a duck goose?

jamie,
@jamie@zomglol.wtf avatar

@ramsey @josh In case it helps, I provided an alternative a few years back

https://zomglol.wtf/@jamie/109288152821638814

jamie, to random
@jamie@zomglol.wtf avatar

I started out today with 2 meetings on my calendar. So far I've had 7 of them so now I've only got 3 left.

Word problems in elementary/primary school math didn't prepare me for this.

thomasfuchs, to random
@thomasfuchs@hachyderm.io avatar

I know weird opinion, but if it’s “Chrome only” it’s not the Web.

It’s Flash, but worse.

jamie,
@jamie@zomglol.wtf avatar

@thomasfuchs There was a period where I'd hear people say something was "a web standard" as soon as it landed in Chrome, even if it was still in early stages.

I don't hear it much these days, but I've also been less exposed to front-end things the past few years, so I don't know if it's still as prevalent.

veganstraightedge, to random
@veganstraightedge@ruby.social avatar

Re-upping my plea for help!

I’m really jammed up on this and blocked by it

When I get it solved, I get to delete 55 triggers and a gem!

And then a similar amount in three other rails apps

Help!?
https://ruby.social/@veganstraightedge/112340159120060018

jamie,
@jamie@zomglol.wtf avatar

@jesseplusplus @veganstraightedge Was just coming to say this. I don’t see a lot of (or any?) functionality in ActiveRecord for dealing with triggers or SQL functions. The raw SQL with the IF EXISTS clause seems to be the best route.

jamie, to random
@jamie@zomglol.wtf avatar

LLVM stands for "Ladies Love Virtual Machines"

nyquildotorg, to random
@nyquildotorg@fedia.social avatar

I thought TFW meant "that face when" for decades, because every single one of them I ever saw featured a gif of someone making a face.

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg TIL that's not what it means!

davetron5000, to random
@davetron5000@ruby.social avatar

Feels like all these new Browser APIs can be used to make an app or make it easier to do so but will instead be used to juice engagement, show ads, beg for mailing list signups, and generally make for a very poor user experience on web sites that just render text.

jamie,
@jamie@zomglol.wtf avatar

@davetron5000 This is the primary reason I thought Apple was doing the right thing by not adding Web Push and related APIs to iOS Safari, regardless of whether their reasoning was altruistic (it probably wasn't). I was actually frustrated that they caved.

It reminds me of a tweet I saw when Edge adopted Blink that said something like "it's wild that the only thing preventing a browser-engine monopoly is Apple's bullshit Safari policy on the iOS App Store".

nyquildotorg, to random
@nyquildotorg@fedia.social avatar

hot take: having Matrix as a dependency is one of the worst parts of Beeper's architecture. Reminds me of the sage biblical advice about not building your house upon the sand.

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg Matrix as in the chat server?

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg Huh, that’s unexpected

paris, to random
@paris@hachyderm.io avatar

massive shoutout to the #kubecon photographer for taking a shot of me where i don’t have my mouth open (yes i talk and laugh a lot don’t judge)

https://www.flickr.com/photos/143247548@N03/53610651135/in/album-72177720315561784/

jamie,
@jamie@zomglol.wtf avatar

@paris RESPECTFULLY 👀

camertron, to random
@camertron@ruby.social avatar

People say Ruby is slow, and sure, it can be. But the amount of time I spend waiting on JavaScript tools absolutely dwarfs the time I spend waiting on Ruby tools.

By at least an order of magnitude.

jamie,
@jamie@zomglol.wtf avatar

@codefolio @Schneems @camertron This thread really makes me wish one of my talk proposals for RailsConf this year hadn’t been declined. That talk shows that pure Ruby code can be faster than C extensions in some situations, even without a JIT. I actually forgot to enable the JIT for it. 😄

The C extension used in the talk was Hiredis, and I was going to show a pure-Ruby implementation of Redis (just a couple commands) that used 2-5x less CPU time than Hiredis for values of all sizes.

recursive, (edited ) to random
@recursive@hachyderm.io avatar

I was thinking about how some people think knowing SQL databases is "I can recite all the different kinds of joins", but 2014-me is here to tell you that knowing databases is also deep knowledge about what happens when you're throwing tons of data in there and deleting data and auto-vacuuming has no hope of keeping up and it all slows down to shit. And it's on a customer's system and you can't merely throw more hardware at the problem. :)

jamie,
@jamie@zomglol.wtf avatar

@recursive I was scheduled to give a talk about this at a Neo4j conference a while back! I was going to be comparing and contrasting how an RDBMS (Postgres, specifically) and Neo4j traverse your data on disk during queries.

Unfortunately, that conference was in April of 2020 so it was canceled due to the pandemic. 😭

https://graphconnect2020.sched.com/event/ZjEC/contrasting-neo4j-vs-rdbms-a-performance-deep-dive

nyquildotorg, (edited ) to random
@nyquildotorg@fedia.social avatar

Me in the early 2000s: "stop abusing country code domains! .tv is not for television. .fm is not for radio."

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg TIL both of these things. I knew about .io and a couple others, but I think I just associated .tv and .fm with TV and radio because I started seeing them more around the same time .biz, .name, and the rest of that generation of TLDs were announced. Assumed it was all part of the same group.

nyquildotorg, (edited ) to random
@nyquildotorg@fedia.social avatar

Tracy Chapman getting accolades again after a country artist covered her song makes me think of Anne Preven, from the band Ednaswap, who wrote a song that would later be covered as a one-hit-wonder by Natalie Imbruglia.

I believe she did pretty well royalty-wise, but it's gotta smart seeing someone else get all the glory from your hard work.

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg This is part of why I think Tracy and Luke's performance together at the Grammys was such a big deal. The entire time he was looking at her with deference, the way a golden retriever looks at its human. He was very explicitly following her lead.

And the crowd cheering their lungs out for her, and only for her, also sent the message that everyone knows it's her song. Luke was just marketing it.

Not a perfect situation, for sure, but easily the best of its kind that I've seen.

evan, (edited ) to random
@evan@cosocial.ca avatar

"Fediverse of Things"

jamie,
@jamie@zomglol.wtf avatar

@evan Reminds me of this thing I wrote a while back

https://zomglol.wtf/@jamie/109294979207786797

jamie, to random
@jamie@zomglol.wtf avatar

I bought the face computer

bonus: dog butt photobomb

nyquildotorg, to random
@nyquildotorg@fedia.social avatar

[opening fridge, rooting around inside] "we got soda, the purple stuff, Sunny D, lead, asbestos, lead-sbestos..."

jamie,
@jamie@zomglol.wtf avatar

@nyquildotorg I can see and hear this toot. At least the first half of it.

mhoye, to random
@mhoye@mastodon.social avatar

By 2026 Covid policy is going to be “Smell this Tabasco and if you can’t you can go home half an our early as long as you work through lunch”, and you pay for the Tabasco.

jamie,
@jamie@zomglol.wtf avatar

@mhoye
> The American military fundamentally relies on a reliable supply of 18-24 year olds without better options

This tracks. Out of the 55 men in my platoon in basic training (20+ years ago), 50 of us said we were there for benefits and college money. It was our only option for either one.

Fitting that I write this while actively nursing a condition that the VA refused to acknowledge as service-connected. 🙃

jamie, to PostgreSQL
@jamie@zomglol.wtf avatar

I learned more about GPT-based AI algorithms from this article than from literally everything else I've read about it so far combined. It implements a GPT algorithm using #PostgreSQL.

https://explainextended.com/2023/12/31/happy-new-year-15/

jamie,
@jamie@zomglol.wtf avatar

My favorite part: "How do we decide which properties these vectors represent? We don't. We just provide enough vector space for every token and hope that the model during its training phase will populate these dimensions with something meaningful. GPT2 uses 768 dimensions for its vectors. There is no telling in advance (and, actually, even in the retrospective) what property of the word will, say, the dimension 247 encode. Surely it would encode something, but it's not easy to tell what it is."

jamie,
@jamie@zomglol.wtf avatar

The implication seems to be that the people who invented the algorithm (and those who use it) have literally no idea how the data will be structured in the model and have no control over it.

They just tweaked some math until it spat out words that sounded good for the corpus and the input. And this is what companies are betting everything on right now. 😬

jamie,
@jamie@zomglol.wtf avatar

@recursive It's like someone heard about the infinite-monkey theorem and was like "I bet we can do that even faster with computers instead of monkeys"

jamie,
@jamie@zomglol.wtf avatar

@recursive It also makes me think a lot about this quote from 1864, attributed to Charles Babbage:

"On two occasions I have been asked, — 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' … I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question."

160 years later, people aren't even asking the question. They're just assuming the answer is "yes". 😭

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • kavyap
  • DreamBathrooms
  • InstantRegret
  • magazineikmin
  • osvaldo12
  • mdbf
  • Youngstown
  • cisconetworking
  • slotface
  • rosin
  • thenastyranch
  • ngwrru68w68
  • khanakhh
  • megavids
  • ethstaker
  • tacticalgear
  • modclub
  • cubers
  • Leos
  • everett
  • GTA5RPClips
  • Durango
  • anitta
  • normalnudes
  • provamag3
  • tester
  • lostlight
  • All magazines
    •