Seems reasonable. Thank you for the response.

https://linuxiac.com/netbsd-new-policy-prohibits-usage-of-ai-code/, probably.

AI wasn't used for writing code. But its used for two things

1. Generating a large number of malformed PDFs that break the compression
2. As a second judge to compare the compression result, as going through 12k pdfs by myself was infeasible. However I did go through 2k of them by myself

Currently no. But thank you for the recommendation. I will add it

Yes, I actually have been working on implementing it. I have been able to get it working, just not in the browser yet

the rise of AI has taught me that simply not using computers is way more viable than i thought.

I literally learned to crochet because i feel existential dread and need more practical skills so i'm not at risk of computers suddenly becoming unusable and my life being in tatters.. Being able to make things is actually so fucking comforting, it's a great way to pass time too.

yeah no sorry, you don't just get to conveniently ignore all the systematic problems with AI usage.

AI datacentres are displacing people, deforesting huges swathes of land, placing massive loads on electrical grids which makes electricity more expensive and encourages more fossil fuel usage, using shitloads of water for cooling; and AI is replacing workers, sneakily influencing users, and people offloading cognitive effort onto it results in their skills atrophying and code becoming unmaintainable.

Using AI to write code is like always buying fast food: it's bad for you, bad for the environment, fucking expensive, and means you never learn how to actually cook your own food.

How so?

not just evangelism but the general weird culture around it, i feel like it's gotten partially absorbed into the crypto/linkedin sphere where people just constantly post technobabble-filled affirmations to show they're part of the group.

ai usage?

Sounds like another human brain, hired and paid legally and with consent. Should tick all your boxes besides free. You gotta break ethics for free.

You're welcome. I will check out Audile.

Not sure, it is downstream from NewPipe so I'm not sure how much vibe coding would even be necessary. But I haven't checked.

@SocialistVibes01 OK, thanks. Maybe you can let Imnebuddy have this link: reddit.com/r/freebsd/comments/CoffeeBSD ― an AI-free fork of FreeBSD

Maybe your instance has defederated from .ml since then.

https://lemmy.ml/instances

We're those pesky commies after all...

Thanks so much! Librivox, Image Toolbox and audire are a fantastic discovery!

Btw, I just found Audile which is more feature-rich and developed more actively than audire.

WizeStream (YouTube, PeerTube, and SoundCloud client)

Is it vibe-coded? The sync feature is a dream come true, but I usually don't use vibe-coded projects

I don't need to edit PDFs very often but I happily host BentoPDF just in case.

But a major feature I need is signing with a CAC. I noticed you said something in the past that it may not be feasible?

@SocialistVibes01 I can't see any of the three previous replies in Mastodon, is that normal? A limitation of Lemmy, maybe?In Lemmy (not logged in), I see a response from Imnebuddy less than thirty minutes ago. beige.party/@Imnebuddy@lemmy.ml (logged in to Mastodon) shows nothing more recent than lemmy.ml/post/41935087 – ten months ago.#Fediverse #Lemmy #Mastodon

So originally this comment suggested an alternative to FreeBSD that takes an Anti-AI stance called CoffeeBSD. It's specifically a fork of FreeBSD 13.0, which is a bit odd since the untainted commits to FreeBSD occur in later versions. It removed the lua bootloader interpreter and added a few md files and hasn't received updates in a few months. I suggested this as I thought it would be interesting to keep an eye on, but as for now it seems yet another fork with not a lot of support and could likely die off. It's created by a contributor of the AI-less fork of vim, EVi, as well as other projects.

35% of those polled want AI-assisted code to be disclosed while 22% felt it should be treated just like any other code contribution and only 16% want AI generated code to be rejected outright.

I find it depressing that many developers don't want to even bother taking the time and care to disclose when AI assisted some code contributions. I'm tired of most software eating its own tail and not taking precautions in case legal issues occur which would lead to reverting the AI assisted commits.

I use bentopdf for a few months now, it's great thx for making it. One thing I wanted to ask is, is there a way to convert multiple images into separate PDF files in one go? When I add multiples I get a single file of them merged

So freebsd doesn't even want me to consider them? Good to know, perhaps they should rename themselves to fraibsd.

Thanks a lot! I am glad.

yonks is a new term I learnt today lol

I've been using BentoPDF for yonks, and have been deploying it for clients too.

Thanks for your work!

Nope, Hyper keeps alt text, tags and reading order intact at every compression level. It's made to produce conformant PDFs by default unless you choose otherwise

I haven’t dug into it yet, but does pdf compression strip screen readers’ ability to read the pdf? Like if I have alt text embedded in a pdf for a screen reader to be able to read, does it strip that out?

Asking because I try to keep ADA compliance in mind, regarding things like text on images. PDFs are great for keeping a page formatted exactly as intended… But they’re fucking awful from an accessibility standpoint, because screen readers typically can’t interpret them.

The go-to solution is to embed alt text that describes what is on the pdf. Because a screen reader can interpret alt text. But lots of sites (and I’m assuming compression methods) seem to strip this, for some reason.

Thank you very much!