Yeah, forgejo will give you many of the features of GitHub. Not the proprietary ones like the Actions Marketplace of course, but a lot of equivalent features. It’s lightweight enough though that even if you never use it for anything beyond git, creating pull requests, and some basic CI, it’s not going to require much power to run it.

Do you need the public to have access to it? That would be the only reason for federation that I could think of. That’s not available quite yet.

I self-host forgejo, it’s one of the easiest systems I self-host.

But which features other than a plain git repo are you looking for? That will mostly determine your options. There are tons of git repos, and even just a plain git repo on a server with an ssh tunnel is enough if you don’t need anything beyond that.

I’m just talking pure cost to manufacture for each. The cost of the hardware is higher: faster CPU, faster GPU, additional RAM, additional storage, higher end cameras, etc. That is where the cost of the phone comes from, so you can’t compare cost of a Fair Phone to this one any more than you can compare the cost of a Pixel 9 to the Pixel 9a. Both have basically the same software, warranty, parts availability, etc., but the 9a was about half the price of the 9 because the “a” series is a lower-end phone overall.

But the specs aren’t comparable. The Murena phone is a higher-end device than the Fairphone. And the Fairphone does get incentives/subsidies for the Android version which is one of a few reasons the Android version is cheaper than the version sold with /e/OS.

The cost of the device is not being subsidized by all of the bloatware installation income, selling of personal info, etc. I think a lot of people don’t realize that inflation really shot through the roof on the last decade and it’s only because products are subsidized by the sale of your data and other unethical practices that product prices didn’t go up. Food uses shrink-flation to hide inflation, tech uses selling your personal info.

Depends on what you want. You can have the application have an https certificate which could either be one issued my a globally trusted issuer or could just be a self issued certificate that caddy is configured to trust. And caddy can then add the globally trusted certificates from let’s encrypt or whatever. But that definitely requires extra steps. Just, how secure do you want to be?

I use forgejo on a raspberry pi.

Don’t include the non-encoded part of the data or it will corrupt the decryption. The decoder can’t tell the difference between data that’s not encoded and data that is encoded since it’s all text.

I also switched to the Transit app for public transit after google maps stopped showing the option to start navigation and only allows for searching for routes now. Not sure why it stopped working. Maybe because I use GrapheneOS, but it worked on Graphene before with no issue, so could be a combination of policy changes and using a non Google OS or something. It works fine with driving and walking still.

For driving, I still can’t find a comparable app that has both real-time traffic which is essential in my city to avoid constantly fluctuating bottlenecks due to construction, traffic pattern changes, and rush hour traffic. I found one at one point, but it didn’t read the names of streets, just said “turn right” instead of “turn right on first avenue”. This is confusing when streets are close together or it’s treating alleyways as streets, etc. CoMaps is the best I’ve found if it had traffic routing.

Do you mean this config option?

[server] 
hosts = 0.0.0.0:5232, [::]:5232

That is binding the service to a network interface and port. For example your computer probably has a loopback interface and an Ethernet interface and WiFi interface. And you can bind to an IPv4 and or IPv6 address on those interfaces. Which ones do you want radicale to listen to traffic from and on what port? The example above listens on all interfaced both IPv4 and IPv6 and uses port 5323 on all. Of course that port must not be in use on any interface. Generally using this notation is insecure, but fine for testing. Put the real IP addresses when you’re ready.

Yeah, the definitions are actually more about alignment with the US political parties rather than left or right. And since both parties are demonstrably right of center, just to different degrees, the bias meter should only be used to determine which political party’s sponsors likely biased the article.

For example, an article saying climate change is not human caused and presenting debunked evidence will be ranked mostly center and second mostly right. But an article calling for incentives to reduce use of fossils fuels will be ranked mostly left. That’s mostly center if anything. An article calling for the government to explicitly force companies to stop using fossil fuels would be mostly left and center. One further advocating for the government to take over energy companies that don’t comply and make energy production public would be mostly left. Just presenting scientific evidence and refusing to give a voice to debunked “alternative facts” is not a leftist position, it’s a centrist one at best and should be the baseline.

Yeah, it’s easy enough to configure it properly, I have it set up on all of my servers and my laptop to treat it as a network mount, not a local one, and to try to connect on boot, but not require it. But it took me a while to understand what it was doing to even look for a solution. So, hopefully that saves you time. 🙂

NFS is really good inside a LAN, just use 4.x (preferably 4.2) which is quite a bit better than 2.x/3.x. It makes file sharing super easy, does good caching and efficient sync. I use it for almost all of my Docker and Kubernetes clusters to allow files to be hosted on a NAS and sync the files among the cluster. NFS is great at keeping servers on a LAN or tight WAN in sync in near real time.

What it isn’t is a backup system or a periodic sync application and it’s often when people try to use it that way that they get frustrated. It isn’t going to be as efficient in the cloud if the servers are widely spaced across the internet. Sync things to a central location like a NAS with NFS and then backups or syncs across wider WANs and the internet should be done with other tech that is better with periodic, larger, slower transactions for applications that can tolerate being out of sync for short periods.

The only real problem I often see in the real world is Windows and Samba (sometimes referred to as CIFS) shares trying to sync the same files as NFS shares because Windows doesn’t support NFS out of the box and so file locking doesn’t work properly. Samba/CIFS has some advantages like user authentication tied to active directory out of the box as well as working out of the box on Windows (although older windows doesn’t support versions of Samba that are secure), so if I need to give a user access to log into a share from within a LAN (or over VPN) from any device to manually pull files, I use that instead. But for my own machines I just set up NFS clients to sync.

One caveat is if you’re using this for workstations or other devices that frequently reboot and/or need to be used offline from the LAN. Either don’t mount the shares on boot, or take the time to set it up properly. By default I see a lot of people get frustrated that it takes a long time to boot because the mount is set as a prerequisite for completing the boot with the way some guides tell you to set it up. It’s not an NFS issue; it’s more of a grub and systemd (or most equivalents) being a pain to configure properly and boot systems making the default assumption that a mount that’s configured on boot is necessary for the boot to complete.

I have m.2 hats for the couple of raspberry pis that need more intense disk operations. Never use SD cards or flash drives, which generally end up being just SD cards in a USB package.

Copyright in general should be strictly limited to an extremely short time, like maybe 1-5 years. After that others should be allowed to use and expand on it unless you release a new work that expands on it yourself. Trademarks eliminate the confusion about who published it and if you aren’t actively using the content, it should be given to society to benefit everyone. This would promote progress and competition. Extended copyright, especially, is only useful for people and companies who don’t want to be productive and just get paid for one thing their ancestors/predecessors did ages ago. The original design for copyright said exactly this would happen.

Problem is many of us are stuck with very low upstream bandwidth due to cable company ISP monopolies and/or data caps or just were running things on a small raspberry pi or something and the malicious requests will create extra expense or flat put denial of service for real traffic.

To a point yes, for the crawler bots, but Anubis uses a lot more resources to keep the bots busy than a simple firewall ignoring the request. And if there’s no response vs a negative response, the requests are likely to fall off more quickly. And the even more significant load might be from malicious login attempts which use even more resources and Anubis likely won’t be as effective on those more targeted attacks depending on the types of services we’re talking about. Either way, firewall blocks are way, way less resource intensive than any of that, so as soon as you open up that firewall and start responding to those malicious or abusive requests they will become progressively more resource intensive to mitigate.

I think it’s safe to say that all of the LLMs have been training their systems on any site they can get their hands on for some time. That’s why apps like Anubis exist trying to keep their crawlers from killing their bandwidth since LLM companies have decided to ignore robots.txt, copyrights, licenses, and other standard practices.

You can mitigate some risks with software like fail2ban to slow down some of the hacking attempts, but you will still be susceptible to, sometimes unintentional, denial of service attacks from ever persistent “AI” crawler bots as well as the constant barrage of automated hacking attempts. If you’re bandwidth is not able to handle it or you have bandwidth caps, you’re likely going to have issues.

Are you wanting something that you don’t have to download from GitHub yourself (so a project that hosts a docker container somewhere and just code is in GitHub is OK), or are you looking to boycott any project that is not boycotting GitHub and so any part of that project should not use GitHub for any code at all in which case possibly even dependencies should not be on GitHub even if they publish their distributions elsewhere? Or somewhere in between?