This is a bit depressing.

https://www.nerdmeritbadges.com/products/octocat

This underlying problem needs to become a priority. Either pressuring #github to go free software or getting free software to go elsewhere.

Github becoming synonymous with open source just muddies free software waters more.

Free Software needs free tools.

@satchmoz In the long term people will cry in disappointment when the company eventually tanks or gets bought and corrupted. But something else will replace it. Alternatives will be able to step in at the right moment.

Meanwhile, grown-up open source projects which have existed much longer mostly still run their own infrastructure, so it's _not_ going to brielfy take out absolutely everything.

@satchmoz BTW I am always amused when a new project arrives at the Apache Software Foundation and demands that their primary repository will be on Github and the ASF tells them "No, that would be incredibly short-sighted; you can only have a mirror there". This discussion repeats roughly every 6 months...

@stsp I don’t see what is “incredibly short-sighted”. It is inexpensive, reliable, and effective, no? What is the thing that these naïve people don’t get? If this is so obvious and these arguments are made so often, feel free to point me to some blog somewhere. I’ll read.

@paco The point is to stay independent. ASF projects do not host critical services outside of ASF infrastructure. This way the ASF can ensure long-term stability.

This is not specific to github and it applies to any critical services (of which version control is just one).

@stsp @paco

https://www.gnu.org/software/repo-criteria.html

@stsp @paco

I realize that's going to draw fire from both sides since it's GNU.

Would be happy to see an extended critique of proprietary code hosting from, say, an OpenBSD perspective.

@deejoe I also think you need to define “proprietary code hosting”. I asked naively before (literally because I don’t know). What is “proprietary” about GitHub? Is there some licensed extension to the git protocol? Is there something they require of a project or an individual that is “proprietary”? I thought git was git, and they’re effectively a commercial, value-added git-as-a-service business. I’m open to being corrected.

@paco

Can you fork it?

If no, it's proprietary.

@deejoe That’s an extraordinary extreme position. At some layer in the stack there will be “proprietary” software. That free OS runs on CPUs that run proprietary microcode. The network routers (my ISPs, if not mine) that connect the hosts to the net run a proprietary OS. Look at the OSI reference model. How many layers can one operate with zero proprietary software? Maybe layer 5 on up? Proprietary layer 7 code is bad, but we shrug at layers 1-4?

@paco

You've shifted the discussion.

You started with "define proprietary".

Now you're arguing *for* proprietary.

Which, you know, fine: You be you.

But let's just be clear the direction you've taken this.

@deejoe I asked for the definition because I didn't understand what was "proprietary". One pays to use their service. That's true. But that's neither here nor there for the definition, right? They don't make all the code available that they write/run in order to deliver the service one pays for. That seems to be the sticking point. The API that they support is open (right?), and they receive no rights to your IP if you host code there, right?

@paco

https://mastodon.sdf.org/@deejoe/99933207332221161

If the criteria in the links in the above do not apply, it's proprietary.

@deejoe I've read these criteria and I understand them. But what drew me into this conversation was the assertion that "GitHub"—just the use of it as a service to host code—was incompatible with these principles. Do they impose license conditions on code they host? Does using GitHub to host one's code somehow tie your hands on how free and open you can make your code? I'm trying to get down to the root of the "GitHub == proprietary" assertion. If my code is on GitHub is it less free?

@paco

It's an unfree dependency.

I think it's healthy to talk about how much of a priority it is to remove that dependency, but unhealthy to deny it exists or that it has particular consequences.

@deejoe We're talking "free" WRT rights, not cost. And using GitHub doesn't grant anybody any rights and it doesn't restrict which rights I can grant. I don't see how it is an unfree dependency. Does signing up at GitHub cause users to agree to so something that we don't want them to be forced to agree to? They're just an online service. I'm expecting a pointer to some Ts&Cs or license language at GitHub where you say "THIS is what people should not have to agree to, just to host their code"

@paco

I'm sorry to put this so bluntly, because I'm pretty sure you understand this, but ... you realize Github runs *software* to provide that service, right? And that when you use Github, you are using that software?

That's the essential point: It's software that isn't free.

@deejoe Perfect. That's where I thought you were heading, and I think it's an impractical stance. The CPU your free OS runs on has proprietary microcode at its core. The hard drives and RAID controllers you might use all have proprietary firmware/microcode. Routers, switches in your network? Most have proprietary microcode/firmware. How does one run a service equivalent to GitHub in speed, reliability, scale, etc. but having no proprietary code in the infrastructure?