There is no magic place that will promise to host your code forever, for free. No trustworthy organization or company has that responsibility, or is going to claim they will fulfill it.
If you promote self-hosting, which has a cost, consider you are now really promoting an alternative to *paid* accounts with commercial git hosts.... not their free offerings.
There are so many recent knee-jerk reactions / suggestions of alternatives to GitHub, GitLab, and similar situations that don't take into account that stuff actually costs money to run & host.
Bouncing between free hosting options is a process that's more or less guaranteed to have no end.
We need to be seeking out, or creating, stable *paid for* options, which do not have the unwanted baggage of VC funded or big-corp owned platforms.
@dctrud good points.
I have wondered if you could provide a public service with a backend composed of volunteers' home computers composed over a VPN. All computers are partial replicas but with enough full replication when combined with peers, even sharing load they can serve with only the data they have.
This would let the cheapest compute hosting to be used to host the public endpoint (hosting bandwidth use might be double, to route from VPN nodes to clients).
@dctrud For the future of donation-supported Git hosts, look at donation-supported fedi hosts... done just up and fold when the owner burns out.
Donation-supported needs a charter and continuity plan. Done way of knowing when the current owner burns out, the service will change hands gracefully.
@CarlCravens indeed - that's a big issue with donation support.
It's difficult to establish a properly binding charter, continuity plan etc. because it requires real expertise, time, and money to do so. Unless donations are sufficient to pay someone to to that stuff, and carry out management type tasks periodically, then it's a very big ask.
@CarlCravens Looking at Codeberg - it has a lot of structure in place. I'm not sure if anyone is being paid to maintain that structure though.... and I'm not sure I consider it really a stable project unless the donations cover that.
They say "As of July 2020, with all expenses frozen, we have a runway of ~12 years, so you don't have to worry that our service will suddenly disappear."
But... how can the expenses be frozen? Are people doing a lot of work for free that they'll tire of?
@dctrud Living in the free tier is basically siphoning off a company's marketing budget indefinitely. Someday it's going to dry up or they'll decide it's better spent on people who might someday make a purchase.
@dctrud Of course, the problem is, you can't dodge big-corps. If you sign up with a service not owned by a big-corp today, it might be tomorrow.
I was quite fond of my Nest thermostat and my Fitbit. And chose them for being independent products at the time. Thankfully I never bought a Roomba...
Ultimately self-hosting or community hosting is the only stable paid-for option long-term.
@ocdtrekkie I don't think this is entirely true. There are web hosts such as nearlyfreespeech.net who have been around for ages, and are very unlikely to be bought. For git, SourceHut (which definitely wouldn't suit everyone) is a company, but I can't see that being acquired by big-corp.
I suspect a large percentage of non VC funded, non exponential growth focused companies never get acquired.
There *should* be the ability to be a business and offer this stuff.
@dctrud I mean, it depends how successful it gets. GitHub was bought. I wouldn't be shocked if GitLab gets acquired someday, it is moving in the direction to, and I always felt like Google would consider it someday potentially to hedge against Microsoft's GitHub ownership.
SourceHut probably won't be acquired but only because Drew is aggressively opinionated enough on making his projects extremely not corporate friendly in ways that will also probably limit their overall success.
@dctrud I'd say the top three or four companies in any given market are subject to risk of acquisition. And they're usually the top because they're the best.
So if you want independent companies that aren't likely to get acquired you need to examine the field and avoid all the best options...
@ocdtrekkie so isn't it a case of supporting a range of non hyper-growth businesses then? Businesses that can cover their costs, make a profit, but aren't trying to become big enough at all costs that they become interesting for big-corp to acquire?
We need to have an ability to arrive at a point where successful != as big as possible.
@dctrud Now you're just asking people to be intentionally less ideal though. Generally either you want more money, or you want more users, or both, or if you think your product is the best solution, you want it to reach as many people as possible for their benefit... but then it's also appealing for acquisition.
Maybe we just need companies run by people like... religiously opposed to ever being acquired?
@ocdtrekkie I mean personally I don't think there are naturally just 3 or 4 "best" companies/products in area X.
How many different cars can you buy? How many different furniture companies are there?
I think in software / SaaS everything has been distorted by huge VC / corp money that leads to most money = most free stuff = best... and a lack of room for solutions that genuinely would be better in function for other sets of users.
@ocdtrekkie anyway... It's a big issue with no real solution. My original point really is that self hosting and co-ops simply aren't going to make a dent... and have their own significant risks in terms of longevity etc.
@matt it's a great archive. But how do you git clone from it? I believe softwareheritage.org is a more focused archive project for code.
@matt we've gotten to the point now where tooling expects to pull stuff from git repos. Where history of the code is expected to browse easily. Downloading a tar file of source from archive.org just isn't going to be a substitute for a GitHub or similar.
@dctrud Ah. Sounds like your tooling is really holding you back. Hope you can find something more flexible
A quiet single user instance for dctrud.