Additional note on Codeberg, which I think is great as a project, but I got curious on what infrastructure they are running on and how reliable this would be for larger corporate repos.
[...] Although aged, the performance (and even energy efficiency) is often not much worse than with new hardware that we could afford. In the interest of saving embodied carbon emissions from hardware manufacturing, we believe that used hardware is the more sustainable path.
[...] We are investigating how broken Apple laptops could be repurposed into CI runners. After all, automated CI usage doesn't depend on the same factors that human beings depend on when using a computer (functioning screen, speakers, keyboard, battery, etc.). If you own a broken M1/M2 device or know someone who does, and believe that it is not worth a conventional repair, we would be happy to receive your hardware donation and give it a try!
[...] While it usually holds up nicely, we see sudden drop in performance every few days. It can usually be "fixed" with a simple restart of Forgejo to clear the backlog of queries.
Gives both early-Google as well as hackerspace vibes, which can or can not be a good thing.
Their reliability is not great unfortunately. Currently their 24h uptime is 89% for the main site. They are partially degraded right now.
The 14 day uptime is 98% but I think that’s actually because some of their auxiliary systems have great uptime, the main site is never that great it seems.
Yeah, they were down last week too. It's hard to run an open git forge on a small volunteer team, the workload is read and write heavy with endless "customers" (or bots).
To be fair, Codeberg isn’t for corporate repos, it’s for FLOSS projects. Take a look at their Terms of Use. They don’t aim to be a commercial provider, rather the opposite.
oh wow I had a larger cluster than that since I was 20 more than half a decade ago, considering that the costs appear to be so low maybe I should also pop out few free services since at the moment I pay $600+ just on power costs alone for idle hardware on my personal cluster. If anyone has any ideas feel free to email me at: news.ycombinator.com.reassure132@passmail.net
I hate these constant drama posts, but I am all for seeing competition. I think it's good to have a couple of top-tier companies offering the same service, and especially with git, it's been... lacklustre outside of Github, I'd say. Bitbucket was totally nice, but Atlassian and Jira and meh... Github has (mostly) steered clear of cross-product promotions until the CoPilot era washed all over us, and I wonder for how long they can continue to thrive off the power of brand-awareness.
Same effect at play watching all the top-tier AI corps under heavy competitive fire still, trying hard to keep the audience attached while battling to stay on top of (or keep up with) the competition. This mainly (for now) benefits the user. If OpenAI were to trailblaze on their own, we'd all be paying through the roof for even the most basic GPT by now.
I always (very naively, I fully get it) wonder if someone at GitHub could take a minute and check the logs (if there are any at this level) from a week ago or so and scan them for patterns? The code seems to grab a few files off of GitHub, use Github actions, etc. -- perhaps there's a pattern in there that shows the attacker experimenting and preparing for this? I assume most people at this level have VPNs and so forth, but I'd never underestimate the amount of bad luck even those folks can have. Would be interesting, I know I'd have a look, if those logs existed.
Amazing, I wonder what their interview process is like, probably whiteboarding a next-gen LLM in WASM, meanwhile, their entire website goes down with us-east-1... I mean.
Careful: NPM _says_ they're up (https://status.npmjs.org/) but I am seeing a lot of packages not updating and npm install taking forever or never finishing. So hold off deploying now if you're dependent on that.
They've acknowledged an issue now on the status page. For me at least, it's completely down, package installation straight up doesn't work. Thankfully current work project uses a pull-through mirror that allows us to continue working.
This morning I was looking to maybe replace my Macbook Pro 2018, which had the horrible keyboard and finally seems to be crippled enough to not be fun to use anymore — now this!
However, I have been disappointed by Apple too many times (they wouldn't replace my keyboard despite their highly-flamed design-faux-pas, had to replace the battery twice by now, etc.)
Two years ago I finally stopped replacing their expensive external keyboards, which I used to buy once a year or every other (due to broken key-hinges) and have been so incredibly positively surprised by getting used to the MX Keys now. Much better built, incredible mileage for the price. Plus, I can easily switch and use them on my Windows PC, too.
So, about the Macbook — if I were to switch mobile computing over to Windows, what can I replace it with? My main machine is still a Mac Mini M2 Pro, which is perfect value/price. I like the Surface as a concept (replacable keyboards are a fantastic idea, battery however, super iffy nonsense), and I've got a Surface Pro 6 around, but it's essentially the same gloss-premium I don't need for my use.
Are there any much-cheaper but somewhat comparable laptops (12h+ battery, 1 TB disk, 16-32GB RAM, 2k+ Display) with reasonable build quality? Does bypassing the inherent premium of all the Apple gloss open up any useful options? Or is Apple actually providing the best value here?
Would love to hear from non-Surface, non-Thinkpad (I love it, but) folks who've got some recommendations for sub $1k laptops.
Not my main machine, but something I take along train rides, or when going to clients, or sometimes working offsite for a day.
thanks for the hint, spec-wise, this is exactly what I meant, 1tb ssd, 16gb ram, 16 hours of battery, very nice. then I saw it's 1700 EUR where I am at the moment, so pretty much Macbook Pro price :(
Nov 22, 2025 https://blog.codeberg.org/letter-from-codeberg-onwards-and-u...
Quotes from their website:
Infrastructure status [...] We are running on 3 servers, one Gigabyte and 2 Dell servers (R730 and R740).
Here's their current hardware: https://codeberg.org/Codeberg-Infrastructure/meta/src/branch...
[...] Although aged, the performance (and even energy efficiency) is often not much worse than with new hardware that we could afford. In the interest of saving embodied carbon emissions from hardware manufacturing, we believe that used hardware is the more sustainable path.
[...] We are investigating how broken Apple laptops could be repurposed into CI runners. After all, automated CI usage doesn't depend on the same factors that human beings depend on when using a computer (functioning screen, speakers, keyboard, battery, etc.). If you own a broken M1/M2 device or know someone who does, and believe that it is not worth a conventional repair, we would be happy to receive your hardware donation and give it a try!
[...] While it usually holds up nicely, we see sudden drop in performance every few days. It can usually be "fixed" with a simple restart of Forgejo to clear the backlog of queries.
Gives both early-Google as well as hackerspace vibes, which can or can not be a good thing.