Arguably the first branch is backwards. I don’t use windows because I fear technology and strongly distrust Microsoft.

Probably in 2025, we at Meta, as well as the other companies that are basically working on this, are going to have an AI that can effectively be a sort of midlevel engineer that you have at your company that can write code.

~ Mark Zuckerberg, Jan 2025 (source)

If you have a kindle you can hack it and load PDFs onto it. The koreader is better anyway.

I don’t think this will work within the context of a feed of posts. You would have to make at least 1 additional comment for every post in the feed to fetch the comments for a post. So if you fetch a feed of 50 posts, you will have to make 51 requests. If a post has too many comments to fetch in one page, you will have to iterate through all the pages until you have all the comments. So it’s actually >=51 requests. Though I suspect you could get a good idea of a posts comments by fetching just the first page of comments.

PieFed seems to have tags, but I’m not exactly sure how they work. But that might be a better place to start.

But where do you store the computed tag? I guess you could hack it by having a bot that comments the computed post tags on the post itself, but that’s messy.

There are also many Lemmy instances that are intentionally blocking VPNs because they have to to stay afloat.

Idk, but a lot of Lemmy instances also don’t play nice with VPNs. Some of them are using Cloudflare, but idk if all the ones blocking VPNs are the ones using Cloudflare. But bot traffic is a big problem, and Cloudflare poses a solution to that. It’s not the ony solution, but it is a pretty good one.

You can’t really have a free social media network and not have it block VPNs. At least not with large public instances.

I wouldn’t know. I use neovim btw. No ai features for this guy

I love federation, but why does a git server need federation? Am I missing something, or is this federating for the sake of federating?

Not that you should vibe code, but you could vibe code this so easily. Have it output a static website. Give the source code a scan if you’re paranoid. Check the network tab if you’re really really paranoid. But literally you could have it output this as a static index.html file that you drop into your browser of choice.

This is the only type of coding LLMs should ever be used for imo. A small, very clearly defined task that is very easy to verify if it works. And code that won’t infect a larger project.

Edit: as others pointed out, that url isn’t base64 encoded. You would have to clearly define what you are trying to do if you want this to work. For example, do all urls follow the same format as the above?

I’m not an apple hater, but that’s kinda insane considering how hostile they are towards developers.

Yes! There would be a spec written for blocklists that could be applied to Lemmy or PoeFed. You can quickly toggle on/off a blocklist. For example, I can subscribe to a “US Politics” blocklist, and toggle it on/off when I need a break from news.

I wonder how that works. The point of password hashing is to uniquely scramble your password. So userOneHash(“password”) should give a different output than userTwoHash(“password”) even if they use the same password. So your password manager shouldn’t really be able to generate the same password hash since an infinite number of hashes can be generated from the same password.

Can you also use a list of common passwords and a ruleset you apply to those common passwords, and then hash(applyRule(commonPassword), salt) == compromised hash ?

I actually didn’t realize pepper was a thing. I mostly do frontend. But that’s really interesting!

But if you use a salt that is global to your site/server, you still have this problem: If a hacker cracks “p@ssword” in your database, they immediately know all users that also use “p@ssword”. Imo the biggest benefit of using salts is two users with the same password get different hashes. Right?

I’m not saying using a global salt isn’t better than no salt, but I do think you’re missing out on a huge benefit of using a per hash salt. Keep in mind I’m a frontend engineer not backend or security lol.

even if someone brute forces an offline copy of the hashes they wouldn’t result in actual useable passwords

I thought you were suggesting that salted hashed passwords were uncrackable but maybe I misunderstood this

Edit: I understand the offline pepper part now. My bad

I don’t think that’s how salts work. I might be wrong, but I think it works like this

Password + Salt -> Hash

• “p@ssword” + “hakf” -> “hifbskjf”
• “p@ssword” + “jkjh” -> “gaidjshj”
• “p@ssword” + “afgd” -> “afgdufj”

Notice how those 3 users use the same password, but the different salts results in 3 different hashes. That doesn’t make it any harder to crack a single hash, but it means I have to crack the same password 3 times. It just slows down password cracking.

Edit: my explanation isn’t wrong, but I didn’t understand the pepper part until now. So I understand the above now.

But if you can solve the hash by generating password guesses, hashing them, and comparing them to the hashed passwords in the database. Say I hash “p@ssword” and I use the salts sorted in the stolen database. I find that jon@example.com uses “p@ssword”. I then go to Amazon.com, login with Jon’s account, and order a bunch of stuff to my address.

Salt just makes it so I can’t hash “p@ssword” once and find everyone with that password the database. Instead I have to hash “p@ssword” multiple times. It really only slows me down.

I’m not a security expert, can someone tell me if I got that right?