Also, export your DBs first, and snapshot the export instead of the raw DB files

borg

I once watched a depression era family member of mine dismantle a Chinese take out container, flatten and wash the cardboard part of it and then put it in the paper recycling, and straighten out the wire handle to put into a little jar with other wires and bits of metal.

Same family member once showed me a stack of paper grocery bags she had taken into her place from her building’s paper recycling, upset that someone had thrown away perfectly good paper bags.

I think 8 hours starts to get into territory where they might get an informational message about the delay? That also starts to be long enough that the emails might get lost in the distant past in the client and never be seen, by the time they arrive.

I think when I used to do this, it was one advisory message every 24 hours that a message was holding in the queue, and after 5 days it would bounce, but I have to assume that those limits have shrunk in the modern day. How much, IDK; it might be worth experimenting with it though before committing to creating that situation since it might not go okay.

SMTP is designed with queues and retries

Unless something has changed massively since I was deeply involved with this stuff, the people that sent you email may get a notification after some hours that their message is being delayed, and maybe after like 24-48 hours they might get a bounce. But if it’s just your SMTP server going down for an hour or two every now and then, the system should be able handle that seamlessly (barring some hiccups like messages showing up with timestamps hours in the past which sometimes is confusing).

You’re the only one talking sense and you are sitting here with your 2 upvotes

The AI company business model is 100% unsustainable. It’s hard to say when they will get sick of hemorrhaging money by giving away this stuff more or less for free, but it might be soon. That’s totally separate from any legal issues that might come up. If you care about this stuff, learning about doing it locally and having a self hosted solution in place might not be a bad idea.

But upgrading anything aside from your GPU+VRAM is a pure and unfettered waste of money in that endeavor.

You’re going to think I am joking but I am not. Multiple people have sworn to me that this works for a common failure mode of HDD drives and I’ve literally never heard someone say they tried it and it failed. I’ve never tried it. Buyer beware. Don’t blame me if you fuck up your drive / your computer it’s connected to / anything else even worse by doing this:

1. Stick it in the freezer for a short while.
2. Take it out.
3. Boot it up.
4. If it works, get all the data off it as quick as you can.

Corporation: Yo can you add a button so I can see all my employees’ screenshots? And maybe get like a little report of what % of the day they’re spending on doing exactly what they’re told? And then like an automated email to HR and their manager if it drops below a threshold…

Ask GPT to rewrite your configuration, check over it with diff to make sure it didn’t do something dumb, bingo bango

Stop, stop. It hurts, it’s too real.

Linux router + iptables + wireshark for debugging maybe? Esp if things are "just not working” sometimes, using simpler and more transparent tools so you can dig into what exactly isn’t working seems like it’d help

Prediction: It’s shit

Tor’s obfs4 protocol is pretty difficult to block, and it has some other transports that are options if obfs4 is unusable in a heavy censorship regime. This page is a good overview of how to start; with the right transport and bridge setup it’ll be extremely difficult for your ISP to prevent you having access.

You could make your home server a securely-accessed onion site and connect to a remote-access-via-web service you’re running there. That part might be a little challenging (and this process overall may be overkill) but it’d be very challenging for them to block it, I think, so if you’ve tried some things and had no luck, that might be the way to do it.

Be careful obviously

Honestly having GPT write one-off code for you for particular selected pieces (esp ones that require a lot of domain knowledge) works pretty well in my experience

Personally I like upvotes/downvotes. Both are pretty useful information to have and those are the two numbers that my brain translates most cleanly into seeing what’s going on.

Also I want to put in my perennial pitch for the software to display a popup or something the first time you vote, saying hey warning your votes are not private on any federated software! Don’t vote with an expectation of privacy.

Does it work out okay with 12 cores purely on CPU? About how fast is the interaction?

I played around a little with Ollama and gpt4all but it seemed to me like it wasn’t fast enough to be useful on pure CPU, but if I could just throw cores at it then I might revisit the issue.

I thought everyone else just did what I do – if there’s a squiggle, take away the squiggle part. If something’s missing, make a blank line and then blindly bounce on the tab key until Copilot fixes it.

That’s step 1, and if that doesn’t work, step 2 is to actually look at what’s going on and try to fix it.

Yeah agreed. All the people in here talking about signals… you’re not, like, wrong, but you missed the nature of the question i.e. “hey it seems like this GUI application doesn’t deal well with Unix signals, so what else can I do”

My first thought would be to use xdotool. Does that work for you?

Federation in general seems like it’s been flaky as hell ever since a month or two ago… my guess (especially if you see it on some Lemmy servers and not others) is that it’s delayed and will show up eventually.

It’s possible that there’s some unexpected behavior that leads it to federate some places and not others, but my expectation is that once your comment is on lemmy.ml’s server, it’s federation delay between the Lemmy servers and nothing to do with you specifically.