One of the best things ever about LLMs is how you can give them absolute bullshit textual garbage and they can parse it with a huge level of accuracy.

Some random chunks of html tables, output a csv and convert those values from imperial to metric.

Fragments of a python script and ask it to finish the function and create a readme to explain the purpose of the function. And while it’s at it recreate the missing functions.

Copy paste of a multilingual website with tons of formatting and spelling errors. Ask it to fix it. Boom done.

Of course, the problem here is that developers can no longer clean their inputs as well and are encouraged to send that crappy input straight along to the LLM for processing.

There’s definitely going to be a whole new wave of injection style attacks where people figure out how to reverse engineer AI company magic.

Anybody seriously believing this has a misunderstanding of how little people care about what OS they use and how much they care that it works the way they expect.

Because after taking a quick look at that first or second page, I don’t even go back. I just head to another search engine 😅

By rewarding mysterious “quality content” indicators that SEOs know how to game with shit people absolutely do not perceive as quality.

My solution is a bit old school: A raspberry pi connected to my network and running miniDLNA. It has an externally powered USB hard drive. My TV runs Android and I have VLC installed. Any DLNA client works including Xbox and mobile phone apps too.

I don’t think mini DLNA is even updated anymore so eventually my solution might stop working but it’s been running solid for 10 years

Oh and about 40% of your codebase will be xml files with magic strings that at runtime magically call a python function. Because fuck you.

Hahaha… This is pretty much spot on about the whole back end

Agree with another commenter that documentation is an issue. If you’re a developer, it’s not too bad. It uses a kind of homebrew model/view/template python backend that might baffle you for a while.

If you don’t need to modify it or integrate it too much with your bank, e-commerce sites, cash registers, etc then it pretty much works out of the box.

Odoo came from a fully open sourced project years ago and is getting more and more closed and expensive over time as they go upmarket. If you really want to own your data I’m not sure that’s a good direction.

There is a fork called Flectra that is usually about a version behind Odoo but reintegrates a lot of the paid features of Odoo for free, but its documentation and community is even smaller.

That overly aggressive issue manager closing tickets because the ticket opener didnt reply fast enough

“Can’t replicate, closed.”

The guy was a the senior software dev at his first startup. Not sure if he’s written a line of code since then, but he’s at least spent some time in the trade

Possibly to run those strangely shaped outdoor billboard signs

Yes. I want to put it into cloud storage that I control and can move around as required, so that I can’t run out of space. And so that I don’t need to manage a local physical drive.

The only thing holding me back with Immich has been the non-obvious back end storage options.

I’d like to keep my data in an S3 bucket, but from what I can tell it’s still unsupported because of some limitations with fuse.

It’s been pretty stable and usable.

But has there been an increase in hiveminded threads and trolling?

At first it felt like dissenting opinions would lead to an informative discussion, but more and more having an unpopular opinion - or just stepping into the wrong thread - means someone resorting to insulting your mom. Ok, thanks Xbox live gamer.

For sure. If you really need to host thrn get the right package of internet services for that activity. If you need a fast fix for a small issue, this method gets around most ISP attempts to block hosting.

Haven’t done any comparisons, but I’m using this in cases where my throughput isn’t too important. I can stream my webcam over it in HD, establish an rdp connection, ssh back through, and control web interfaces like octoprint. It’s acceptable.

The big advantage is that a $2/month vps gets a static IP, therefore I don’t have to worry about a VPN provider changing the IP or blocking ports.

Tunneling an encrypted connection through an encrypted connection is pretty much always a bad idea but when you’re evading CGnat or other network blocks, sometimes it’s the best we can conjure up on a whim.

🫤 that does suck. Probably so they can charge more for a hosted package?

For reestablishing the connection, use the wrapper ‘autossh’. It can be run from systemd so that it’ll auto start and restart as you command it.

I have a couple instances using this that are absolutely rock solid after years.

A solid workaround is an ssh reverse tunnel with gateway ports enabled. You can do it for pennies with a cheap VPS.

With this option you open an ssh tunnel outbound and then you can connect back through it from the other side for whatever local services you want to run.

You first need a VPS with a public IP. Here’s a guide that explains it: https://www.howtogeek.com/428413/what-is-reverse-ssh-tunneling-and-how-to-use-it/

Just remember to enable gateway ports in the VPS side sshd.conf and disable or adjust any firewall on the VPS so the internet can come in through the VPS ip address and tunnel back to your local system.

This. I had the same situation being put behind CGnat and told them my security webcam needs port forwarding from outside and they had me back to a public IP within minutes.

I remember using Trillian to overcome the MSN/AIM/ICQ boondoggle and it didn’t take long until all of those hosts started to fight back by complicating their protocols making it hard for the Trillian devs to emulate authentic clients.

The truth is that Youtube doesn’t want creators to own their identities because Youtube wants to own the viewers and tell them to watch whatever will make Youtube more money. Kind of the same as why reddit believes it has a moral authority to take over and control a subreddit that was built by a moderator.

Youtube wants viewers to be “Youtubers” not “Mark Rober viewers” or whatever. Otherwise Youtube becomes some kind of free hosting service. But they CAN help new creators get discovered by vast quantities of viewers if they so choose, which they offer hypothetically in return for a piece of the ad revenue which they can secure when the viewers are kept as Youtubers.

Youtube will work hard to break any 3rd party front end such as this one for sure.

Anyways, I hope GrayJay can attract a good following before the platforms figure out how to block them.

The first team or company who figures out how to let content creators own their identities completely - ActivityPub style on their own instances - combined with someone who figures out a compensation/revenue system and a way to drive viewers will probably be the Youtube/Twitch killer.

Actually, now that I think about it, what’s the core difference between everyone spinning up a WordPress that’s RSS fed into a reader?