In that case I can really highly recommend it. Nixos on the server is fantastic anyways, and the only hurdle to recommending simple-nixos-mailserver is that most people are not familiar with nix… 😄

It’s a bit unconventional maybe, but I vote simple-nixos-mailserver - IF you are curious / willing to learn nix. It’s essentially just sanely configured dovecot, postfix, rspamd.

My config for those three combined is about 15 lines, and I have never had an issue with them. Slap on another 5-10 lines for Roundcube as a webmail client.

Since it’s Nix, everything is declarative, so should SOMETHING happen to the server, you can be up and running again super quickly, with the exact same setup.

Yep, that’s right. In theory you could share the encrypted DB with the public and not degrade security. (Still don’t do that though…)

Huh. I update my revanced YouTube app every 6-9 months

We were talking about SwiftKey

Who knows?

Unless a piece of software is open source, you cannot know.

Fail2ban allows you set different actions for different infringements, as well as multiple ones. So in addition to being put in a “local” jail, the offending IP also gets added to the cloudflare rules (? Is that what its called?) via their API. It’s a premade action called “cloudflare-token-multi”

A high-quality laptop without any branding.

I’m currently using a 9-year-old, woefully underpowered laptop made by Xiaomi. Full aluminium unibody, and NO logo. Not printed on, not etched in, not glistening only in the right light. NO LOGO.

I’m not a billboard. I’m not responsible for your brand recognition. Ironically though, far more people have come up to me and asked “hey, what laptop is that” than ever would have cared if there was a logo on it.

It also just looks and feels fantastic, all-aluminium-no-logo just looks so sleek.

So yeah. I will not be upgrading until I find another laptop of the same build quality, with no logo. Tuxedo has that option for most of their laptops, but for some reason not for their only current full-aluminium body -.-

Oh, and don’t come at me with stickers.

I switched a couple of months ago, from SwiftKey. Had been using that for ever, long before Microsoft bought it.

NGL, the transition was a bit rough, and the first month my error rate spiked. All good now though, plus Futo has a bunch of super useful features SK never had. Overall, very happy.

We expose about a dozen services to the open web. Haven’t bothered with something like Authentik yet, just strong passwords.

We use a solid OPNSense Firewall config with rather fine-grained permissions to allow/forbid traffic to the respective VMs, between the VMs, between VMs and the NAS, and so on.

We also have a wireguard tunnel to home for all the services that don’t need to be available on the internet publicly. That one also allows access to the management interface of the firewall.

In OPNSense, you get quite good logging capabilities, should you suspect someone is trying to gain access, you’ll be able to read it from there.

I am also considering setting up Prometheus and Grafana for all our services, which could point out some anomalies, though that would not be the main usecase.

Lastly, I also have a server at a hoster for some stuff that is not practical to host at home. The hoster provided a very rudimentary firewall, so I’m using that to only open necessary ports, and then Fail2Ban to insta-ban IPs for a week on the first offense. Have also set it up so they get banned on Cloudflare’s side, so before another malicious request ever reaches me.

Have not had any issues, ever.

No. I am not saying that to put man and machine in two boxes. I am saying that because it is a huge difference, and yes, a practical one.

An LLM can talk about a topic for however long you wish, but it does not know what it is talking about, it has no understanding or concept of the topic. And that shines through the instance you hit a spot where training data was lacking and it starts hallucinating. LLMs have “read” an unimaginable amount of texts on computer science, and yet as soon as I ask something that is niche, it spouts bullshit. Not it’s fault, it’s not lying; it’s just doing what it always does, putting statistically likely token after statistically liken token, only in this case, the training data was insufficient.

But it does not understand or know that either; it just keeps talking. I go “that is absolutely not right, remember that <…> is <…,>” and whether or not what I said was true, it will go "Yes, you are right! I see now, <continues to hallucinate> ".

There’s no ghost in the machine. Just fancy text prediction.

Yeah, with seniors it’s even more clear how little LMs can help.

I feel you on the AI tools being pushed thing. My company is too small to have a dedicated team for something like that, buuuut… As of last week, we’re wasting resources on an internal server hosting Deepseek on absurd hardware. Like, far more capable than our prod server.

Oh, an we pride ourselves on being soooo environmentally friendly 😊🎉

Makes me feel warm around the heart to hear that it’s not just me 🫠

Even with LMs supposedly specialising in the areas that I am knowledgable (but by no means an expert) in, it’s the same. Drill down even slightly beyond surface-level, and it’s either plain wrong, or halucinated when not immediately disprovable.

And why wouldn’t it be? These things do not possess knowledge, they possess the ability to generate texts about things we’d like them to be knowledgable in, and that is a crucial difference.

I’m a programmer as well. When ChatGPT & Co initially came out, I was pretty excited tbh and attempted to integrate it into my workflow, which kinda worked-ish? But was also a lot of me being amazed by the novelty, and forgiving of the shortcomings.

Did not take me long to phase them out again though. (And no, it’s not the models I used; I have tried again now and then with the new, supposedly perfect-for-programming models, same results). The only edgecase where they are generally useful (to me at least) are simple tasks that I have some general knowledge of (to double theck the LM’s work) but not have any interest in learning anything further than I already know. Which does occur here and there, but rarely.

For everything else programming-related, it’s flat out shit.I do not beleive they are a time saver for even moderately difficult programs. Bu the time you’ve run around in enough circles, explaining “now, this does not do what you say it does”, “that’s the same wring answer you gave me two responses ago”, “you have hallucinated that function”, and found out the framework in use dropped that general structure in version 5, you may as well do it yourself, and actually learn how to do it at the same time.

For work, I eventually found that it took me longer to describe the business logic (and do the above dance) than to just… do the work. I also have more confidence in the code, and understand it completely.

In terms of programming aids, a linter, formatter and LSP are, IMHO, a million times more useful than any LM.

LMs give the appearance of understanding, but as soon as you try to use them for anything that you actually are knowledgable in, the facade crumbles.

Even for repetitive tasks, you have to do a lot of manual checking to ensure they did not start hallucinating half way through.

I am using both and this somehow made it to my phone, wtaf

When I first switched to nix, I made an error copy-pasting my hashed password into a secrets file.

Reninstalled the system 5 times, each time immediately locking myself out, almost

Managing ~35 machines without issues now though.

Dang, oof

FWIW, Lidarr works the worst out of the arr stack for me too. I don’t know if there’s just not enough well indexed material in my sources or what, but yeah, not great.

If your entire experience with the arr stack has been Lidarr so far, give it another shot! Sonarr and Radarr work absolutely perfectly. It’s just such a nice feeling to open Jellyfin (or I guess Plex) on the TV and go “oh nice new episode is out!”