@maddy @seabass It would be, although even NetBSD requires at least a 486 now (I'm not sure what the last version to run on a 386 is)

@maddy @seabass No, I'm referring to this kind of thing: https://amazon.com/dp/B00AQT2LRK

It's intended for laptops where there may not be room for an adapter

@maddy @seabass Did you know that you can get PATA SSDs? (but unless it's in a laptop it'd probably be cheaper to use an adapter)

There's an Am386SX in my grandma's house that my dad bought for her and she has never used, I have no idea if it still works.

@seabass congratulations!

@ari I want to avoid requiring non-free software at all costs, so if it was more of an issue I’d probably use something like sethrawall (SSH-based authentication) or just require manual confirmation

@algernon @ari Consider enabling signin via framagit.org (or some other GitLab CE instance that doesn't require non-free software to register). Forgejo as an oauth2 provider doesn't support scopes yet so by signing in via Codeberg gives the forge you're signing in to full access to your Codeberg account, and I'd rather avoid GitHub.

@ari I have the built-in (free software) forgejo CAPTCHA enabled and require email confirmation. Then I just manually clean up all accounts with unconfirmed emails (which is all of the spam accounts) occasionally.

0.0027
https://www.keithcirkel.co.uk/whats-my-jnd/?r=AQwhKP__7-fw

I did switch between two monitors though, that probably helped

RE: https://ck.catwithaclari.net/notes/ajojt88um1yd01wi

@elly I wouldn’t call any of that old, it’s newer than every computer I have.

My main laptop is a ThinkPad T440p (~13 years old, with i7-4700MQ, 16GB DDR3L, and a 500GB SATA SSD (Samsung 860 EVO that’s at least 6 years old although not in constant use during that time)), and I would compile Linux on it.

I do have some newer computers but they’re still all older than 2020:

• My school laptop (and also the one I took to FOSDEM): ThinkPad T480s (~8 years old, i5-8250U, 16GB DDR4, 240GB NVMe)
• What’s going to replace my current server (and I currently use to build absurdly huge stuff like OpenBMC): Supermicro X11SSH-LN4F (motherboard from 2017, CPU from 2019 (i3-9350KF), 32GB ECC DDR4, 2x 1TB NVMe SSDs in btrfs RAID-1 (WD Red SN700, the only new component))

Both of the working smartphones I have (PinePhone and OnePlus 8T) are from 2020. I don’t see why anyone would need a faster SoC than the Snapdragon 865 on a phone, it’s mainly restricted by the shitty operating system (Android). I think part of the reason people upgrade phones so often is a combination of phone manufacturers only providing a few years of updates and difficult-to-replace batteries.

(Also it must be LPDDR3 not DDR3L in your KOHAKU, that SoC doesn’t support DDR3L)

are there any other cases of linux man pages documenting non-existent functions that are not provided by any library and exist only as implementations provided in the man page itself, or is this the only one?

RE: https://berkeley.edu.pl/objects/ccc44049-4f9f-46a6-8ae3-0f98cdf024bb

@ariadne @pius I was reading the string_copying(7) man page to remind me of what the different string copying functions do, and apparently it recommends using strtcpy and stpecpy instead of strlcpy and strlcat, both of which appear to have just been made up by the authors of that man page since it also says "This function is not provided by any library; see EXAMPLES for a reference implementation."

@cwebber this is why it's always important to know where your towel is

@maddy @Chloe surprisingly my server survived a christineboost, but before I installed go-away, scrapers scraping forgejo were enough to slow it down to the point SSH'ing into the host apparently took several minutes

it's currently running on a VM with 4 Xeon E5-1620 v2 (ivy bridge) cores, 8GB RAM, and 200GB HDD storage

@unnick Probably 80/443 or 53. Although on my school network it's actually everything but 80/443 (they have some sort of MitM proxy running on those ports but everything else is uncensored)

@harlow @thing yes, zlib has the same issue. it's probably fine, I'm not aware of it actually being enforced in the context of freely distributable software (but I'm also not australian or a lawyer)

@thing @harlow of course, but that doesn't mean I can't ask them to use another license (which they can of course refuse to do)

(and my criticism did actually result in the requirement to report bugs on unmodified versions upstream being removed)

@thing @harlow as long as they don't accept AI-generated changes to the code you could fork and remove/revert the changelog

@thing @harlow if it’s just the changelog then it seems fine to continue using, just don’t read the changelog and use git log instead

@harlow @thing please use a free software license rather than that self-written proprietary license, there are several problems with it:

• the restriction on using it for AI training is a restriction of freedom 0 (and probably pointless since AI training is usually just ignoring copyright and claiming it’s fair use anyway, and fair use cannot be restricted by a copyright license)
• the requirement to report bugs in unmodified code to the original author also restricts freedom 0 and fails the DFSG island test (what if you’re on a desert island with no internet connection and can’t report bugs?)
• not having the license in the same repo makes it unclear which version applies, if you ever update it, or if someone finds the hfetch repo separately from the one with the license
• your warranty disclaimer doesn’t have “to the extent permitted by applicable law” or similar, which makes it illegal in Australia (although many free software licenses (notably MIT/Expat) also have the same problem, see this for an explanation)