Before GitHub
lucumr.pocoo.org
lucumr.pocoo.org
To me one of the clear things that GitHub gave us was a structure around a person rather than a project. To me it felt liberating to quickly create a repository attached to my name than it was to go through the (what felt to me) very serious process of coming up with a project name and reserving it on sourceforge just to get a cvs or svn repository (along with website, mailing lists, issue tracking(?), etc, etc...). It felt like the mental load of "oh this is just a quick thing" was a lot easier with github.
> It gave projects issue trackers, pull requests, release pages, wikis, organization pages, API access, webhooks, and later CI.
Although it didn't give us this all at once. I still remember when we created a new user account in order to simulate an organisation, before they existed. I distinctly recall discussing with friends if we wanted to set up a bug tracker software for our project with the assumption that "GitHub will probably release one in a few months anyway". In the end we just kept a text file committed in the repository. Issues were announced a few months later.
If I remember correctly, it was also one of the few places sticking to the now-standard passing of the parameters via path rather than the '?' URL query part.
It might not seem like much now, but then the ease and simple beauty of having just github.com/user/repo - not only for web access but also cloning - was definitely some freshness factor.
1. not use query params for key entities in the URL
2. to stick user identifiers at the root path! totally unheard of to occupy such an important path at that point!
Taken together, this was entirely novel. Next to nobody did this. Twitter was the one other notable example, and that's literally all I can think of.
The URL bar was so different back then. It wasn't search by default. The average tech savviness of internet users was higher. People cared about URLs despite the fact most websites had garbage cgi-bin query string slop. Lots of folks had personal domain names. People typed URLs and shared them a lot - so this was a big deal, because they were memorable, unlike the other slop URLs at the time.
To give more character - HN's urls would have been considered exceptionally nice back then. The average URL was way worse and was littered with hundreds of query parameters.
A great deal of websites put your session token in the query params. PHP had first class ways of spending "sessions" to all urls. Essentially a cookie. It was disgusting.
GitHub revolutionized urls as a product. Even today, not many companies followed suit.
They're still the gold standard.
2nd level would be yourname.com.
3rd level would be yourname.site.com, like LiveJournal and Blogspot had a long time ago.
4th level would be site.com/yourname, like Myspace had a long time ago, and Facebook had after that, and Github had after that.
Once you sink all the way down to the obscure depths of 4th level, there's not much difference between site.com/yourname and site.com/whatever/yourname
Very few companies do this. Github and Twitter are some of the only ones. And they were certainly incredibly novel at the time.
Back in the 90's and 2000s, URLs looked disgusting. Remember this [1] nonsense?
[1] https://www.php.net/manual/en/session.idpassing.php (see "example 1")
That may be, but the user I responded to said "top level". Anyways, what looks disgusting about:
yourname.livejournal.com
yourname.blogspot.com
myspace.com/yourname
?
This was all before github, so objectively not novel by then.
The Fediverse makes it much easier to broadcast updates in a truly decentralized fashion. Maybe this can be the discovery layer instead of requiring a centralized social network in GitHub?
Git itself has many of the primitives needed, but issue tracking and CI/CD seem the main pieces still lacking. Seems likely that issues can be solved satisfactorily both natively in git, or in the fediverse. I bet that agents can pass branches/patches amongst themselves just fine, maybe native git (pulling from each others’ remotes on a crontab, or sharing patches by listserv) is actually completely viable now.
I also wonder if agents will cause a shift from forking and contributing back upstream, to more like horizontal gene transfer where people or tight teams broadcast their repos and other projects borrow/steal whatever may be useful. You see this with the Claws where maintainers are simply swamped with contributions and no way to verify whether they are correct/positive. (Things like “first time install on a Mac mini breaks at step 7, due to some iMessage issue” are fiendishly difficult to automate”)
I use Fossil for all my freelance work and it so easily allows me to get right back into the context of a project, niche details and agreements had with a client, etc. No need to pollute the codebase or gather together a million emails or notetaking software just to get back up to speed.
It can still change, I hate the notion that because Git is so culturally embedded we couldn’t ever switch. Fossil makes it super easy to switch and the workflow is actually easier coming from Git.
"Just use SQLite" feels like missing the forest for the tree (the pun being coincidence here). That is, certainly any database out there already use all kinds of very efficient structures to walk graphs under the wood.
Maybe Git has a more bespoke approach to its specific goal of source code as main topic to deal with, and finner layer of abstractions. Which could also explain the clumsy leaky interface it presents.
I don't think he's got public sign ups turned on yet. Maybe hit him up on the Twitter for more info.
[0] - https://x.com/ragelink
"This service is completely free and run because a service like it should exist."
"All public repositories":
https://chiselapp.com/repositories/
I don't know if it has all the GitHub features people may be looking for.
Chisel runs on Flint, "The ISC licensed codebase behind http://chiselapp.com.":
https://chiselapp.com/user/rkeene/repository/flint/index
EDIT: Add "...should exist." sentence from the homepage.
Git certainly had (and perhaps still has) worse user experience, but it worked and felt production-ready, with, of course, one of the largest open source projects in the world using it, and that made all the difference, perception-wise.
It just... never was something majority actually want so they didn't really get any traction.
Issues wise you also get few nasty cases where you really do not want to keep it with project, like having clients send a bunch of screenshots or even videos of triggering some bugs can grow storage pretty quickly... and while extra few GBs on a file server isn't a big deal, keeping it with code repo just so someone can look at tickets locally is PITA, and you quickly get into "let's not use it, it just makes everything complicated and everyone repo bloated".
Someone could probably implement most of fossil features using git as backing store without all that much problems, the wiki/issues/whatever else features would just be separate, parallel branch hierarchy
I explicitly dislike the idea of using Git as the backing store. Forget the fact that Git is not native on Windows and is immensely bloated; the actual .git folder is a mess for backup systems when working locally compared to a single database file.
Sure but there is a big difference between being stored once (modulo backups) on a central server, and every developer needing to download all the resources for every issue and the entire wiki in order to work on the code at all. It works fine for sqlite, because they only have a handful of developers, so it's not a big deal for them each to have their own local copy of everything. But having to download GBs of issue and wiki data in order to make a pull request (however that would work with fossil) or otherwise contribute is a significant barrier to entry.
* Changes being tied to a commit (akin to git notes) so issues could have different status and comments on different branches. No overall view of the project.
* Using their own branches to separate from code state, which made a huge mess when looking at your branches or git history.
* Using a separate space (like git notes does, I forget what this is called), which isn't included in the standard push/pull.
* All of the above being distributed, issues in your checkout could be very different than someone else's, and now you have to merge those too.
* No non-developer UI for project managers to see or comment on issues.
I was exposed to Mercurial before Git and I stubbornly tried to advocate for it over Git for a while. BitBucket, at the time, gave Github a good run for their money and had great Mercurial support and was what I preferred.
I'm not really sure VCS were ever differentiated for there to be a wide world of them. They all solved the same set of problems so similarly that it felt, to me, that there had to be one winner. Right now most of the competition is in the Git Porcelain space.
N.B. I actually have a soft spot for darcs, which was my first actual DVCS. I just loved it so much more than svn and refused to use svn in college and used darcs to actually manage my projects and push them to svn after.
I grew up on CVS and then Subversion. Played with Bazaar a little, mainly because it could use an SFTP location as the back-end.
And I still avoid Git if I can help it. I would/do figure it out when I have to, but it never feels comfortable. Such is my avoidance that I'm dabbling with Jujutsu although I'll still need to really sit down and read through it some more to grok the way it works.
Of course moved on to git but I still think bazaar did many things better.
I was expecting when I make a commit, I would have the facility to specify what issues it addressed and it would close them for me automatically. It seemed there is so much opportunity there to "close the loop" when the issue tracker, etc and integrated in your VCS, but it wasn't taken.
Personally speaking though, I don't want things automagically closed GitHub-style based on parsing a check-in comment. An issue ought to be closed with intention.
Sure, I get that. I was just disappointed that none of the project management stuff seemed terribly integrated in any way from my brief review. It seemed like opportunities there that were not taken.
https://fossil-scm.org/forum/forumpost/e19ed2bfea94fc91f544c...
https://stackoverflow.com/questions/5761669/how-to-fix-a-tic...
Two things I keep coming back to: (1) The "opinionated / small-teams only" critique others have raised in this thread is real, and I think Fossil should own it instead of fighting it. The 5,000-engineer monorepo market is a solved problem (Git won). Fossil should own the 1-50 person bracket — where having issues, wiki, forum, and code in one self-contained, syncable SQLite file is a huge unfair advantage.
(2) AI agents are a brand-new reason to look at Fossil that didn't exist when Git won. Every repo is a queryable SQLite file. An agent reads tickets + wiki + code + history with one SELECT — not 47 GraphQL calls and a rate limit. RAG and MCP setups against the repo become trivial.
We're stuck on the name (fossilforge vs fossilhub). If you have an opinion: https://fossilhub.io | https://fossilforge.io — vote, get early access.
The self-hosted side is already shipping at fossilrepo.io if you'd rather run your own.
--- Disclosure: founder, so grain of salt accordingly.
palaeontology.dev ... too awkward.
museum.dev has a sort of "this is dead" ring to it.
Ironically what fossils are stored in within a museum is referred to as a "repository"..
Well, there's only 2 hard problems in computer science right?
I own a cute domain for that: repositoryum.com. I had the plans for it, but it's one of those that never came to a realization.
Yes there are only 2 hard problems in computer science - caching, naming things, and off by one errors.
- amber.dev
- quarry.sh
You’ll probably need to play with gTLDs to find something that works.
Can also echo “scm” from fossil’s domain:
- amberscm.dev
Along with useX.com, Xhq.com, etc., patterns.
Of the two you have listed, I’d choose fossilforge, but would vote for an alternative TLD since .io has an expected meaning coming from GitHub.
You can even run it on shared hosting as a CGI (never done it myself).
The only thing that took some setup was sending email notifications.
> The "opinionated / small-teams only" critique others have raised in this thread is real, and I think Fossil should own it instead of fighting it.
I agree I use Fossil for multiple personal and small projects. Anywhere I can, really. It is very simple and everything is distributed.
It's another (big) point towards paying someone else to host it.
being a software project makes the environmental connotation less important (using a different project would also cost server energy etc...) so its more clearly tongue in cheek
Could something like github be made with fossil, aka fossilhub?
I believe the answer is ... in theory yes, in practice no. So this already means, if correct, the comparison between git and fossil is incorrect here. Fossilhub would not have dominated; git + github on the other hand did. Again, in theory a fossilhub could win over people to use it (and fossil), but people will compare it to github (back when github was still great) and become quite critical when fossilhub does not offer the same or similar set of functionality. At the least the core functionality - great issues + discussions, easy committing and changing of code and so forth.
Perhaps with enough resources, fossilhub could have conquered the world, but for whatever the reason, it did not, and I think this is in part due to the design. GitHub changed how people interact with repositories. They even made it easy to e. g. add files and change them online, at a later point in time. For instance in one project I am a co-maintainer and I rarely have to use the commandline; I can simply edit via the browser as it is. I don't think fossilhub would have done the same - actually, there is not even a fossilhub, so how would you want to compare git to fossil? It's not just the commandline code. Git has github; while it is a separate project, what does fossil have that people know and use?
> It can still change, I hate the notion that because Git is so culturally embedded we couldn’t ever switch.
We all have our dreams. All desert to become forests or agriculture may be a great idea. Effecting this is hard - but best luck to you betting on fossil here. I don't see it happening. Git raised the barrier here, even if only indirectly via github.
Why? I don’t see any practical reason.
If Git was created by a random dude, it would never taken off.
Hmm that's interesting to hear. I'm starting up a very small business this year (one guy selling shit at a craft fair booth) and I'm using plaintext accounting files in a Git repo for it. I wonder if Fossil could keep me from re-inventing some note taking and record keeping approaches. I'll have to look into what it can do.
Yes, I am someone who, when my branch code-wise is ready to be merged, I will still take however long it needs to clean the history into a coherent, bisectable set of patches.
To be fair, Fossil is kinda cool in many ways, but it's a downgrade for my dev UX. Also the configuration interface for access controls and suchlike to the repo server, issue tracker, etc. is... eccentric. I wouldn't be surprised to find some incorrectly/unsafely configured repos in the wild.
I think this is a bad thing actually. Having something that's centralized but helpful-99%-of-the-time atrophies our collective archival skills. If everything had to be seeded by someone to keep it alive, everyone would be better at holding on to their copies of the things they really cared about instead of being able to assume they can just check it out again when they want to.
There should be no single place that something can be taken down. When a project on GitHub gets DMCAed it takes everyones' forks with it too. Just look at what happened when Nintendo took down the popular Switch emulators in 2024, where archival/continuation efforts consisted of people figuring out who had the latest revision checked out and sharing it. That was only possible because they were very popular project: https://news.ycombinator.com/item?id=40254602
Aside: I really love the Splatoon-ish header/footer animation on this site! Very unintrusive, adds a lot to the vibe, and also quickly gets out of your way as soon as you scroll down. I'm totally going to rip this off lol
The alternative would be many sites, each one of them with their own DMCA rules.
What would be the better alternative?
Still dependent on centralized trackers. Same story with every useful package manager, there's always a middleman.
Also it feels like "if it's not on GitHub, it doesn't exist", which is a bad thing. Feels like too many developers don't know that code can be stored somewhere else.
There is already such an organization in Europe:
https://www.softwareheritage.org/
Underfunded relative to the task, and the (accelerating) speed of free software production.
My laptop is on all the time, make a WASM slave and thousands of developers can give their CPU/Memory/Disk for build slaves.
Like bitcoin mining, there could be some competition between 3 parallel builds to pick the winner if the output is the same.
WASM isn't a magic bullet for sandboxing. CI environments assume a full Linux. So you need to either ran a VM (with the attack surface that implies) or a write an x86 emulator in WASM (which would be very slow).
You also need anti-abuse to stop bitcoin miners from using your system. GitHub probably have full-time employees working on it.
> Like bitcoin mining, there could be some competition between 3 parallel builds to pick the winner if the output is the same.
It's a lot more complicated because many builds are not deterministic, you need to store artefacts, build secrets, etc.
Companies like golem.network or iex.ec have been working on this problem for a decade and they are still not easy to use.
My theory is that we waste lots of resources on CI because it's invisible. If you can hear your fans spin up you might try to optimize more, even if the financial cost is still negligible. Back when every developer built the software and ran the tests on their own machine, slow-building projects were the exception not the norm. There was also a much faster feedback cycle for changes to the build and test processes.
Tooling around a self-hosted CI workflow really needs some work to make it as convenient as even the very difficult GitHub Actions.
Codeberg does have some free CI runners, although I’m not sure what capacity they currently have, and how well it would work out if everybody decides to switch from GitHub today. They do encourage you to pick the smallest runner that works for you, and keep the workflows lean: https://codeberg.org/actions/meta
Or you can self-host your own runners too, of course.
Edit: there’s one caveat – Forgejo Actions are Linux only. If you need Windows or macOS runners, this won’t work for you. But... you could have a readonly GitHub mirror (which you should probably do if you want people to discover your project), and use the GitHub Actions runners for free :-)
If anything Codeberg’s legal structure (being a non-profit) and vision makes it a lot more aligned with the objectives of free and open source projects than GitHub in the long run (which has always been the case but it’s just abundantly clearer today).
I think “for-profit corporations providing high quality public services for free” was a zero interest-rate phenomenon and never sustainable.
Fun fact: Django is still running on Trac today, and has been for more than 20 years now: https://code.djangoproject.com/timeline
(I was not involved in setting that one up, though it's possible I helped get the private Trac that pre-dated it running, I honestly can't remember!)
I guess that award goes to Gitlab now, which I will probably also remember fondly.
Honestly, Gitlab's CI is one of its killer features.
I really enjoy Gitlab CI.
But, nearly everything else (kubernetes management, AST, AI "DUO", work items, milestones, snippets, workspaces, "operations", "security dashboards", "value stream managements", "service desk") - ugh, awful.
I guess some of the artifact repository stuff is nice, but like; their terraform repository is probably the worst of all choices, all the downsides of the HTTP state backend and no upside..
It's so hit and miss; but! the CI is actually good..
I use the command invocation to run python scripts. I tend to believe that CI is just a controlled way to do things you would do locally.
Painfully true - I remember a company I was at replacing GitHub and a bunch of other tools with GitLab because it was better to pay for one tool that does it all. Kind of.
Everything is deeply nested. Many things hard to find. And the UI can be strangely slow or awkward in spots.
I recently worked a job where the client migrated to GitLab from GitHub at my suggestion. And while I think there was on the whole satisfaction with that, it did some with its own frustrations as well.
Employers used GH so I switched but even now I use GH as a dumb git endpoint and do all my build/deploy with docker and shell scripts so switching for me is extremely cheap.
For work stuff I’ll use whatever I’m paid to use if I don’t get to make the call just as it was back in the svn days.
But, my first issue tracker was bugzilla. Setting that up was a bit of a pain, and it didn’t integrate well with anything, but it was very satisfying to see “Zarro Boogs”.
I remember being interested in MantisBT and a few others (Launchpad for BZR?) mainly because it seemed they made a bunch of decisions for me.
In retrospect, it was probably the flexibility of projects like Bugzilla that heralded the "opinionated" approaches to software that followed. In many ways software also follows the patterns of the language they are written in . Bugzilla was written in Perl, so of course there is more than one way to do anything.
I had forgotten about Mantis, but that was the first tracker that the non-programmers in our group were comfortable using. It is a bit funny how quickly we all migrated to Github as a larger community as it became the default for just about everything.
Trac was perfection. Self-hosted and just the right amount of features.
I didn't understand this. I perceive the Zig project and Mitchell Hashimoto / Ghostty to be at similar levels of "weight and signal." Especially because Ghostty is written in Zig.
It feels kind of like saying, "Oh, I didn't take this seriously when it was just Fabrice Bellard, but now that an actual influential person like Guido van Rossum is doing it, it's real."
But you might think about it differently.
GitHub has their own: https://archiveprogram.github.com/
Software Heritage is a non-profit funded by UNESCO: https://www.softwareheritage.org/2019/08/05/saving-and-refer...
Although they're mostly the code / commit history, not so much surrounding metadata like issues, PRs, discussions, wiki, etc.
What Microsoft acquisitions still have any of their original spark left in them? Or Oracle? Or IBM? Or Google? Etc…
Hell, some Microsoft originals from inside the company like Xbox have even lost their edge.
Money is great and I’m sure I’d take the big check, too, but I’m surprised more tech founders don’t think of their legacy in this way when they decide to sell out.
It’s considered a grand accomplishment to essentially lead the wonderful thing you created to its slow demise and hand it over to apathetic quarterly earnings zombies.
And now it’s gone.
(As a fallback, why not email the maintainers instead?)
His post is a response to Mitchell's, from just a few hours ago. I'm impressed with how quickly he wrote a long-form, high-quality, well-reasoned post.
Is there a link to Mitchell's post somewhere? I can't find one in the article.
My memory is telling me that I set up Bugzilla at some point after I joined the core team, though that may have been someone else. When git started becoming a big deal, I spearheaded converting our big SVN repo into many git repositories, and set up the cgit web interface for it. We were still using Bugzilla at that point.
I left the project in 2009 or 2010 or so (I'd joined a small startup and didn't have much time for OSS, sadly), and rejoined in 2022. In the intervening years they'd stood up a GitLab instance with CI runners, and had migrated everything from Bugzilla to GitLab issues.
It's still a very small team (handful of active people), and the infra is mostly managed by one person. It's all very doable, even for small teams. We're very lucky that our infra is generously donated/sponsored, though we also probably get just enough in regular donations that we could pay for it ourselves if we had to. I really appreciate that we're not dependent on Github/Microsoft for anything. Seriously, if you told me 20 years ago that Microsoft of all companies was going to own the largest OSS code forge in the world, I would have thrown up. It still doesn't sit well with me.
Just like many others are actually dependent on IBM, Oracle, NVidia, Intel, AMD, Google,...
Which is understandable when only those provide paychecks, even if they could actually do much more given how much they use.
Nearly any of us could run an XMPP/Matrix server and federate with friends or Nostr/{0xchat,whitenoise}, all with audio, video, text, file exchange etc, yet less than 1% do that.
Simply people, techies as well, have forgot the meaning of personal ownership and therefore are owned by someone else.
This archival project already exists, it is funded by France computer science research agency Inria, by Europe, and maybe by the UN through UNESCO if I'm not mistaken, but I really think it should still receive much more attention and funding to really pursue its goal: Software Heritage.
Then it's BYOR -- bring your own renderer. Trivial CLI bugtrackers, agentic nonsense, pretty web stuff, whatever and the data lives in the repo.
And probably the network/black-hole effect of platforms like GitHub, Linkedin and the like are hard to achieve with fully distributed solutions, all the more when the other side is backed by huge capital which absolutely love concentration of power.
And on top of that, spam is a huge issue. We've progressively further and further locked down new accounts on gitlab.xfce.org because the spam situation has just gotten so bad. We actually don't allow new "native" account creation at this point, and ask people to come to our Matrix channel to ask for an account. We do allow SSO from Github and gitlab.com, but some spam still sneaks through that way too.
I have my own personal Gitea instance that just doesn't allow outside users at all. I'd love to move all my personal projects to it, but at some point I would actually like to try to start a community around one or two of them, and I don't want to have to deal with spam.
Thank you for maintaining Xfce! It's the best de around.
(Edit: Turns out there’s a very obvious and widely used option. git format-patch + git send-email is used to develop major open source software such as Linux, GCC, and Git itself.)
That's fine for pull requests, perhaps, depending on your preferred UX, but what about a bug tracker? CI? Release artifacts?
But that wasn't hip enough so everyone moved to GitHub.
You may prefer email over the latter, but I think these days you're in the minority. Your level of derision at this fact doesn't really change anything.
Before we had FTP, which made easy to mirror stuff with "lftp mirror -p".
HTML is not good for archiving.
We have to rethink those protocols so that mirroring is made easy, and "git clone" is not an answer sorry.
https://wiki.archiveteam.org/index.php/Category:Software_arc...
And, at the same time, we have these AI tools that make it super easy to roll your own version of something. Feels like there's a big push from both sides to start reducing external dependencies.
Thankfully, this doesn’t have to be the case – Forgejo imports pretty much everything mentioned. (1) Whether you decide to move to Codeberg or host your own instance, you won’t have to lose the context.
But I definitely agree we should also have a metadata archive of some sort, for both GitHub, Codeberg, and self-hosted projects.
(1): Not sure about code reviews, and you don’t get the security advisories, though I’m sure it can be replicated with a CI workflow somehow?
I do not agree with details, because it was for me before and after git.
So the hidden denominator here is and still is git, which sparked a tooling frenzy with reversing flow by being online server first (it wasn’t named cloud back then).
So even today, all splinters are doing something around git. That hasn’t changed.
What I really miss is the some sort of standardization that GitHub provided for a brief period of time. Projects would get no love aka stars when you couldn’t easily be used even for the experts. Some convenience as well as tooling evolved, devops became a thing.
I think of the future of a concept called cocooning. The JavaScript expert of today would be puzzled to write code on a notepad in a html file, because it has become so meta, being TypeScript essentially.
There is so much tooling going on that especially Python before AI already felt like I would miss something out if I would code more than 100 lines and that there must be libraries that abstract this all away and instead of coding I should google better.
AI is one thing, but the cluttered tech stacks aren’t really sparking any interest or joy in me, I think it is the not invented here syndrome or because I can story.
I miss the die hard coders, who stick to a tech stack which simply worked, not optimizing for weird use cases which are contrived at worst and rarely needed at best.
This became evident with the decline of data sheets, because Grunt, Gulp etc. as build tools were great but slow. We JavaScript devs couldn’t any longer joke about the compile times of the Backend dudes. And besides that, build times costs you focus, money, cpu time. But this was the main currency.
With AI I stopped trying out lots of tools because they feel like a weekend project by some dude who blasted his Claude budget.
Over are the fork and commit wars. Until AI battles itself this hard for quality source code I will stick to GitHub.
Anyone who is able to just plop a forgejo instance on their own machines... please do that if possible!
Which is often the dream for a small project.
Probably yes, Codeberg are very transparent about their infra details and they don't seem wild:
Specially, I remember not "getting" Github at some point. Bitbucket had mercurial support, sourceforge had SVN, and all the Cool projects lived in SF (I'm talking mid/late 2000s).
The first time I navigated into a github project and just saw the code three I was puzzled. (SF was centered on the project/product while GH focused on the code.
A bigger space isn't necessarily a better one if it's flooded by low-quality content. Does a one-person project on GitHub have any advantage over one that uses a local git repo? I don't think so. It has a disadvantage because you're dependent on GitHub, and even worse, that first-year student learning development now thinks GitHub is an integral part of development.
If projects had to start with a local git repository and a local process, before being able to migrate to GitHub upon gaining minimal notability, developers would be more accustomed to the idea that a project isn't its hosting and that platforms like GitHub are optional. I think this would encourage more local-first attitudes instead of centralization, and that would be good. It wouldn't be seen as weird to send someone a zip file of the thing you're working on.
Plus (only tangentially related) imagine where the JavaScript ecosystem would be if packages like is-array and left-pad couldn't get accepted to npm.
I always found Git to be liberating. I did use a self-hosted BitBucket server, for a while, but now use GitHub.
GitHub gave us pull requests, and a cloud-hosted repo that is managed by others.
I'm quite capable of managing my own repos, but I'm not as good at it as the GH people. If you think their uptime is bad, hold my ginger ale.
But I also don't work on very confidential stuff (anymore). I don't really care if my code gets indexed.
The link leads to a blog post saying how they mercifully unblocked the users in Iran, but that was some time after they first suddenly blocked those users without any recourse and any way to even download their data and private repositories hosted there, without any advance warning…
Tangled supports hosting your own knots. Why wouldn't the current GitHub CEO choose to support self-hosted "databases". GH infra can protect you from all the traffic (ssh and http spam), while you keep the social aspect.
I don't need ~100 extra features, I need a reliable platform. If that means I need to self-host the storage, runners, etc. fine.
> That is one of the great ironies of modern Open Source. The distributed version control system won, and then the world standardized on one enormous centralized service for hosting it.
Cycles everywhere indeed. Perhaps we should ->
> GitHub wrote a remarkable chapter of Open Source, and if that chapter is ending, the next one should learn from it and also from what came before.
Indeed! Try to learn from the inevitable iterations to make the next instance at least that slightly better.
... Where the stuff meets the metastuff it seems all works under very similar forces. My thinking is step-by-step - it works on the individual level, and it scales up.
Day to day is step by step and a step today funds the step tomorrow.
Github might be on the decline, but git != github
Unfortunately I don't have a clever solution (to the social aspect of the problem).
Well ... sad or not sad ...
I remember I was very displeased when Microslop, 'xcuse me, Microsoft assimilated GitHub. But for some time it worked quite ok-ish, to some extent. Only more recently are things suddenly breaking down. I am not sure why they break down right now, but I suspect it has a lot to do with the new AI-focus Microslop pushes onto everything. The AI permeates everything like a virus and contages things (I just tried to create a new word from contagious ...). It seems as if this is the real new corporate identity. As Microslop proceeds to dive deeper into AI (they have no alternative anymore, they already sold their soul to AI), they forget that GitHub used to be about people, first and foremost. Steve Ballmer also had this with his fake antics aka "Developers developer developers" many years ago in 1999 (https://www.youtube.com/watch?v=8fcSviC7cRM - at the least Ballmer was entertaining, the current slop-CEO is boring to no ends).
> So when I think about GitHub’s decline, I also think about what came before it, and what might come after it.
Well - the decline happens largely because Microslop totally struggles. They could have decoupled departments and what not, but they decided the corporate strategy is AI-only now. And this creates a ton of downstream problems. So in many ways the current CEO is to be held responsible; he is in charge since 2014 after all. Fatigue kicks in. I am not saying a single person alone is responsible for failure, but it is clear that whatever the reasons for problems, this comes from Microslop first and foremost. The GitHub team contributes to this decline as well (services no longer working suddenly) but it really is induced from top, aka the overall corporate strategy here. And I also don't see Microslop being able to change course - they overcommitted already, so now the decline is indeed unstoppable.
> GitHub changed how Open Source feels, and later npm and other systems changed how dependencies feel. Put them together and you get a world in which publishing code is almost frictionless
Frictionless is a strange word. I retired from rubygems.org when shopify flooded the zone with ... corporate agenda. The final straw was the 100k download limit ("past that point we disallow you from removing old code you published to rubygems.org", which meant that people would download old code and assume that I would maintain that, which clearly was a lie, so thanks for that RubyCentral ... a year after that they went amok and mass-fired numerous devs; the whole story is a bit more complicated than that, but I can now wisely nod my head, since I retired about a year from that before that mass-fire devs event unfolding). I think if you have a source code hosting service in place, no matter what it is, you need to think about making publishing code super-easy at all times, including the UI. GitHub did this, sort of; I notice this when I compare it to gitlab. Perhaps gitlab has more features, but using it is soooooo much more annoying compared to GitHub. Codeberg lacks features on the other hand. Offering a good service here is actually difficult. It almost seems as if there are no clever UI designers anymore.
> My first Open Source projects lived on infrastructure I ran myself. There was a Trac installation, Subversion repositories
Well. I hated using Trac. Reporting issues is of course possible, but it feels so much more cumbersome than github issues. What could help would be to kind of make semi-universal IDs, e. g. I register once, but then I could use the same account on many different issue trackers. Right now I need to register for each instance and that is just tedious. I keep a password file (don't tell anyone) and I noticed that, say, after about 100 different websites and names and password, it just becomes unmanageable. Yes, I could use software to help me with that, but I decided that I simply no longer want to have a gazillion accounts. I almost never register for phpBB webforums (though discourse appears to be killing phpBB anyway).
> You could find forks, and old issues and discussions all stayed online.
That is true in general, but I would like to remind the blog author here that when the xz backdoors utils were found, Microslop took down the whole repository INCLUDING discussions. I remember that because I also discussed this on the xz utils github issue; and next day when I looked for more discussions, the whole thing was gone. Microslop censored here. Lateron the repository was back again, some days later, but if I recall correctly the whole discussion section was also gone. Microslop did not like the discussion; perhaps it was the author too, but the initial removal was from Microslop. So why was that bad? The whole discussion contained valuable information for people who were not yet familiar with this. Thus, Microslop deprived people of that information. Since then I am very wary about "trusting" Microslop or any private actor here when it comes to censorship. So I would not trust the "discussions stay online" claim here.
Turning and turning in the widening gyre
The falcon cannot hear the falconer;
Things fall apart; the centre cannot hold;
Mere anarchy is loosed upon the world,
The blood-dimmed tide is loosed, and everywhere
The ceremony of innocence is drowned
But as I say, New tech invades the world and makes the perfectly working old tech as incompatible, just by changing the world around it. So git became a necessity imposed.
I'd love to have a longer conversation with you about how we can seed a better system, because on the off chance I'm successful I have a once-in-a-generation opportunity to fix past mistakes.
To be clear for the past five years I've done nothing but write OSS code (https://github.com/conartist6) while sharing pretty much all my engineering thoughts on a public Discord server (https://discord.gg/NfMNyYN6cX), so I'm not very worried at all that a person determined to find out would be unable to tell if I'm legit. You just can't fake 20,000 hours worth of public toil.
The ones who show real curiosity, those are the people I'll give my full attention to any day of the week.
Just as relevant a point…
Not that much smaller right-before GitHub and right-after it became available.
> but in the number of projects most of us could realistically depend on.
Most FOSS I realistically depend on I don't obtain from GitHub actually.
> There were well-known projects, maintained over long periods of time by a comparatively small number of people.
There were even more not-well-known projects, maintained for less time, by a larger number of people. They just weren't that many of them in one place.
> You knew the names.
You absolutely did not know the names. Post author is just thinking of the names they knew as though those were everybody.
> reputation mattered in a very direct way.
And now it doesn't?
> We took pride (and got frustrated) when the Debian folks came and told us our licensing stuff was murky or the copyright headers were not up to snuff, because they packaged things up.
RedHat was just as popular a distribution; and most users used Windows (like they do today); and the BSD distributions were a thing (although we didn't have Apple's BSD, i.e. MacOS)
Bottom line: Inaccurate description of history.
I absolutely knew the names of the people I interacted with and whose projects I used. I even went to conferences with some of them. When I worked on my first web portal for Ubuntu we had a total of about ~4 dependencies and all was vendored. I knew the person who packaged my Python libraries for Debian.
You might call it an inaccurate description of history but it is very much my experience.
I know every name on mysql devel team.
The only reason i subscribe that mail list is: i reported some bugs and need to follows the release.
Signal to noise ratio on those mailing list was high. I can't say the same for github or discord
That's one project. You did not know every name, or most names, in FOSS generally. Today as well - a project which uses GitHub still needs a mailing list, web forum, chat channel (Matrix/IRC/discord/Telegram/whatever) to discuss and coordinate, and that hasn't changed.
As for whether signal to noise ratio on mailing lists was high - that really depends on the list. I don't see that much noise on GitHub repositories, to be honest - it's not easy to post noise that many people will see, so there's not much motivation for it.