The nvim-treesitter repository was archived
133 points by rwdf
133 points by rwdf
This repository was archived by the owner on Apr 3, 2026. It is now read-only.
Anyone know what's up? This is a widely used plugin that a lot of people depend on, would be a shame to see it go.
The maintainer probably got fed up with entitled users: https://github.com/nvim-treesitter/nvim-treesitter/discussions/8627#discussioncomment-16440673.
Wow, telling a maintainer to go away and having them say "ok" and then yank the project is the most intense thing
Love to see it. It’s not like clasom doesn’t have other things to do.
Seriously.
Professor researching MRI reconstruction works on a plugin to his favorite editor only to get told off by a random trend-chasing newbie for not doing enough.
This should either trigger shame and self-reflection among the instigator(s), or they double down with “bragging rights” for what they did. I’m sad it could easily go either way.
Shame and self-reflection is impossible for entitled jerks. They will instead blame the people they attack.
If that guy wasn't so sensitive, he wouldn't have shut his project down! All I did was express my opinion!
The only way to deal with such people is to warn them once, and then block them.
Why are people still on Github at this point?
Simply moving your project to anywhere that requires a login signup screens out 99.9% of the entitled jerks.
And that's before we get into all of the bad behavior of Microsoft.
I think maintainers should be free to use whatever platform they like. There's no need to attack, shame or push your preferences upon them.
Free CI
Free CI with macOS runners
I have a game project I'm working on. I use GitHub Actions to make Windows and macOS builds. Windows is okay, I cross compile from a Linux runner, but I don't wanna figure out cross compiling for macOS. With Actions, I don't have to.
I've considered moving it to Codeberg, but if I did, I'd maintain a GitHub mirror just for the macOS runner.
also filters out 99% of the bug reports you might get - how many of us would create a login on some new platform just to report a bug?
I’ve done it from time to time in the past. Even subscribed to devel mailing lists. I wish I found more opportunities to do so, rather than so much being on GitHub alone.
that's fair. personally, i remember the days of having to create a separate login to interact with every site and as often as not i just didn't. there's a lot i dislike about the centralised and increasingly deteriorating nature of github, but a single login to interact with multiple projects is not one of them.
(i do take the grandparent's point about the low friction also enabling trolls, spammers and drive-by haters in general; it's a hard balancing act. perhaps the happy path is separate sites but something like google auth used in all of them.)
The low-friction with GH also results in the low-quality issues which end up fatiguing maintainers. There should be some system which can ensure that those submittion issues or PRs do their due diligence.
Surprisingly more than you might think. ~90% of the bugs I reported in the past 15 years have not been on GitHub (and I reported plenty). My own projects, which aren't on GitHub, also receive more reports and contributions than any project I had on GitHub in the past (not counting $work projects' company contributions).
My forge has a "Login with Codeberg" and a "Login with GitHub" button, people don't seem to have much trouble spending a minute using them.
if you can log in with github or codeberg that's not the sort of "require a sign up to add friction to screen out the jerks" barrier the gp was talking about though. I do agree with you that centralised authentication but not centralised sites is the best option.
It is, because I can fling the banhammer at them, and they can't come back. It's much harder to do that on GitHub, or with registration, either open or moderated. It's also a step that these jerks don't seem to be willing to take. It does involve Yet Another Account, on a random forge, usually owned by the person they want to bully. That's a very different power balance than on GitHub.
aha, I see your point now! I was thinking of the effort involved in leaving the first entitled screed, not the maintainer's ability to make sure they couldn't leave a second one.
It's not just about the second one, and not just about the effort - it's more about the power balance. On GitHub, both the maintainer and the bully are mere users. The maintainer may have a bit more power, because it's their repo & project. Registering on another platform, even if that's only a "login with github" just be a jerk? That's a considerably bigger commitment to being a jerk. That alone will make many pause and reconsider if they want to sink so low.
On GitHub, being a jerk is sadly more common than it should be, it doesn't stand out. On a person's private forge? Whole different story! It's like... easy to be a jerk on the schoolyard, surrounded by jerk buddies. It's a whole different story going to the other kid's house, knock on the door, and be a jerk to them in their own home.
it is hard to appreciate how extremely difficult unpaid, volunteer open source maintenance is until you start facing this attitude from users.
People would feel bad to talk like this with customer service after experiencing problems with something they paid for. But for whatever reason the same people talk like this when they get something for free.
At someone who helps out with customer service duties from time to time, they absolutely talk to us like this.
fair enough my bad.
But I'm curious, what % of customers actually behave like this? Would you expect it to be a major issue at the lower volumes of incoming communication a small-to-medium open source project will get?
I think it's more about personality than volume. There's a kind of person who thinks "I am very smart and technical and if I couldn't figure it out first try without reading any docs the only response is frustrated violence" and open source projects and my service attract more of this type of person than is perhaps the norm in the population.
There are also very nice and understanding people as well of course.
There are also very nice and understanding people as well of course.
Some times it takes just one asshole to ruin your day.
It is a trope that all the praise in the world doesn't wash away a single negative review.
The percentage is small.
I do support for both Open Source projects, and commercial customers. Which means that over decades, I've run into a lot of these kinds of people.
For Open Source, the solution is to ban them. The problem is more difficult on the commercial side. Many companies make the mistake of seeing revenue, and demanding that their employees put up with all kinds of egregious behavior. It's better to look at the cost of those customers, too.
i.e. "Will my support team quit, or be severely demoralized after working with that customer?"
If so, it's cheaper to get rid of the customer than to replace / train new employees.
The only times that companies accept abusive behavior from customers is either when the corporate culture is abusive, or the employees can be replaced at a moments notice with no training.
This is a real problem and I am not sure how maintainers should handle this problem besides just being quicker with banning . It’s never worth keeping users like this around and always just better hitting the ban button as quickly as possible.
Really sad. I have been a happy nvim-treesitter user for years.
But the maintainer's reaction is understandable. If you happen to read this: thanks so much for the time and work you have put into the project! There are people who really appreciate it.
I'm also grateful. The Scala 3 syntax highlighting story in vim/neovim was truly awful until treesitter and nvim-treesitter.
That was my first experience with it, and today I rely on it for syntax highlighting on any language with LSP server that supports it, which is most of them.
Someone once described open source maintenance as a Benjamin Button situation, where over time your work regresses from "cool rockstar solving previously-unsolved problems" to "janitor answering requests from users who didn't even spend a minute reading the documentation". I think about that comparison a lot.
I think there is two things with Open Source and maintaining things. 1. Open Source software is now more accessible than ever, and with github now making issues and reports is the most accessible than ever. 2. Users, although, important for sure, aren't the most knowledgeable because Open Source software is more accessible.
Although I do think that especially with nvim-treesitter the question isn't so "didn't read the documentation at all", it's different. The maintainer had created a new branch (main) where a rewrite took place with a smaller scale and one which was experimental. So users could unknowingly be on the version they have no idea about. And nvim-treesitter is included in many configurations and not configured by the user themselves. And specifically with this instance the new main rewrite targetted a version which was only released the previous day.
Some of the user comments in the repo are really quite shocking to read honestly. I don't understand how people can be so animated and lack any sense of decency.
It's not that shocking if you've read any thread on here about Systemd or Wayland. People are incredibly entitled to software shared with them for free.
nvim-treesitter wasn't foisted on people using political power and, even if it were, it could be disabled. In addition, as far as I know, nobody from nvim treesitter developers ever tried to claim they were the One True Way(tm), wipe out all the alternatives, and insult end users when bad behavior was pointed out. nvim-treesitter is not a fundamental piece of the operating system that will shut you down if it breaks.
That is a stark contrast to the Wayland developers who have committed all of the aforementioned sins.
The problem with Wayland is that the systemd lesson learned was "We can tell the end users to go fuck themselves and there's nothing they can do about it" rather than "We actually spent 10 years making it do what users asked and everybody eventually adopted it because of that."
See pipewire for contrast in listening to users. In fact, the workaround for the fact that screen sharing STILL doesn't work in Wayland is to use pipewire and dbus(!) to workaround the brokenness. Pipewire could have simply pointed back to Wayland and said "Blame them" but didn't.
That is a stark contrast to the Wayland developers who have committed all of the aforementioned sins.
how/why do people genuinely believe that people (a large proportion of which had spent many years working on x11) spent 10+ years working on wayland primarily, if not exclusively, to screw users (lighter phrasing than you have used)? this cannot possibly be an argument ever held in good faith, i simply don't believe it.
pipewire and dbus(!) to workaround ... Pipewire could have simply pointed back to Wayland and said "Blame them" but didn't.
portals are literally the point and have been in the design documentation for wayland before the first line of code for it was ever written.
i'm begging you to actually read about this instead of falling victim to propagating the opinion of rage-bait youtubers.
how/why do people genuinely believe that people (a large proportion of which had spent many years working on x11) spent 10+ years working on wayland primarily, if not exclusively, to screw users (lighter phrasing than you have used)?
I'm sorry that the Wayland devs tripped over "The first 90% of the project takes 90% of the time. The last 9% of the project also takes 90% of the time. And the final 1% of the project also takes 90% of the time." But they brought this on themselves.
The primary problem was using the political power of RedHat to effectively force Wayland on everybody when it still doesn't work for obvious and not-uncommon use cases. How hard is it to understand that doing such a thing is going to piss people off?
We literally just got a pointer warp protocol within the last 9 months, for example (and still only for certain Wayland implementations and I don't think any of them have hit any of the stable distributions, yet), and I'm pretty sure that feature has been requested for a decade+. That's fundamental to CAD programs and works fine under X11. That's not a small usecase. And there are lots of other people complaining about other features that are also important to their usecases.
When I have to maintain a special X11 installation of Linux to run KiCAD, for example, I'm not "falling victim to propagating the opinion of rage-bait youtubers." When RedHat causes dozens and dozens of projects to get flooded with bug reports because they couldn't be bothered to test their mutter changes against silly programs like, say, KiCAD, that's not opinion that's fact. Telling me otherwise is just going to make me angry.
And part of the flashpoint is the fact that Fedora and Linux have always had a kind of fractious relationship. Fedora wants the benefits of being in the Linux ecosystem while not always adhering to the social responsibilities. And, because they contribute so much code to Linux (I think only Google, Intel and Huawei contribute more), they mostly get away with it.
the driving maintainers of x11 going "actually honestly we can do this better" and going to do so (whether you believe it is better or not) is not a conspiracy. people constantly act like maintaining x11 is free and they're just being anti-foss or anti-you and they're being paid hush money to avoid working on it, but the very few that step up to the plate and attempt to front the cavalry very rapidly discover why nobody else has. "wayland is the future" is because that's where the development effort migrated to, not because it will at some point publicly guillotine x11 in a city square.
The primary problem was using the political power of RedHat to effectively force Wayland on everybody when it still doesn't work for obvious and not-uncommon use cases.
you're again making a really big assumption here that the maintenance burden of the underlying scaffolding here is simply free. the active development effort for both mutter and kwin had moved to wayland and wayland accessories, x11 was (not much right at this time, but had very much begun) increasingly on life-support. gnome was already wanting to default to wayland in 2016 - it worked for them, the maintainers and people actually working on the software.
When RedHat causes dozens and dozens of projects to get flooded with bug reports because they couldn't be bothered to test their mutter changes against silly programs like, say, KiCAD, that's not opinion that's fact. Telling me otherwise is just going to make me angry.
i don't think redhat are the culprits here, and i'm repeating myself a bit, but this is once again expecting that the maintainers continue working on that thing after they have concluded wanting to themselves, so that one of an effectively infinite number of downstream consumers can have a perfect experience migrating.
i can't speak to kicad specifically, but the stable pointer_warp extension was preceded by pointer_constraints_unstable_v1, which is from 2014. the name might look scary but that's just an artifact of how protocols go from rfc to stable in wayland.
wayland being designed in a sort of "capability" mechanism means individual behavioural threads need to be specced out, designed, then spend a bit of time going through an "unstable" period while being refined. yes, it's taking a really long time, and yes it is quite arduous, but the people shepherding wayland as a protocol and maintaining its immediate downstream consumers have collectively agreed it's worth the effort, and that maintaining x11 is not.
And part of the flashpoint is the fact that Fedora and Linux have always had a kind of fractious relationship. Fedora wants the benefits of being in the Linux ecosystem while not always adhering to the social responsibilities.
a significant number of consistent/bigger fedora contributors (i.e., not drive-by or one-time) contribute heavily across the ecosystem with the intent that those changes manifest in fedora as a byproduct of upstream improvements. by design of this they have to adhere to those social responsibilities.
it is of course perfectly understandable and fine to have an altar to wayland in your bedroom, or hate it with all your soul. what i truly do not understand is this fixation that it's possibly history's least efficient coup ever.
gnome was already wanting to default to wayland in 2016 - it worked for them, the maintainers and people actually working on the software.
It doesn't actually "work for them", because they are not dogfoodding. If you look at their screenshots exif, it's always "Photoshop on macos".
If you look at their screenshots exif, it's always "Photoshop on macos".
??? This is like saying they don't dogfood because they play games on windows
Edit: I went and looked at the metadata for the screenshots on release.gnome.org and this doesn't even appear to be true
The alleged rationale here seems really hilarious
I mean, if you really want X11 to continue to be developed, there's this thing in society called "money" that you can use to exchange for goods (such as food and clothing) or services (maintaining a display server). I suggest using money to convince people to maintain it would be a good course of action.
Sadly, we still don't have especially good tools for allowing people to pool money for these things. The main tools I know of are "governments", "companies" and "non-profits", each of which have their own issues. :/
Plus, you know, wealth distribution.
To be fair on a default install of my distro of choice (Debian) getting x11 back is just a drop down box on the login screen. Does it feel weird to have Wayland taking up disk space? Sure. But no one is making me use it
I can easily disable Wayland, or systemd. By just not using systems that ship it. No 'political power' (lol) has ever forced me to use them. The people who work on Wayland worked on X, so the charge of 'wiping out the alternatives' would be very funny there. All the alternative init systems to systemd still exist and really aren't that hard to install if you want.
I actively run several systems that have neither, and they haven't magically stopped working because someone started developing something else and it was adopted by popular Linux distributions. However, thank you for making a perfect display of the exact kind of entitled hysterics I'm referring to!
Exhibit A.
I'd also add Firefox to that list. It's another project that seems attract some of the most vitriolic and entitled responses. I don't always agree with what they're doing, but I'm also very glad I'm not having to put up with what feels like, at times, a weirdly toxic userbase.
As someone that makes security software, entitled users that insist I'm doing things wrong when they disable half of their browser features for "privacy" reasons that break the security software get vitriolic and violent. It's a big part of the reason that Anubis development publicly slowed down a bit last year and I focused on the thing that actually makes me money.
There's quite a difference between raggingn a project you don't like in a public forum and going to the issue tracker for the project itself and telling them to give up...
I'd say another difference here is whether repository being archived is considered the best possible outcome by the complaining people.
I don't know if this is a fault of platforms, but used to be far more liberal back in The Old Days. Piss people off an IRC, get kicked. Come back and do it again? Get kick banned for 24 hours. Do it yet again? We probably won't see you again. Yes, some communities got too controlling, but for the most part online manners were essentially enforced consistly. I can't help but feel ab element of that could do with returning. I think another aspect of this (at least on IRC), was the ability to empower community members to just moderate. I'm not sure GitHub even supports this, but it means there's little incentive to really build communities
As co-moderator of a subreddit where, luckily, the vast majority of people self-moderate, I should say: it's complicated.
Without resorting to moderation: blocking can be very effective – effective in a way that simply could not work in (say) GitHub issues.
Out of curiosity, what do you mean? It's been a long time since I used Github issues, but I don't see why "block abusive user from opening or commenting on tickets" wouldn't work.
It's difficult to explain concisely, or without going wildly off-topic, but the threaded nature of Reddit is a very large part a big picture that includes one-to-one bickering literally falling out of sight without a third person needing to do a thing about the bickering.
Yeah for sure I think moderation tools should be used more, IRC however I think somehow manages to have a general attitude that is sort of condescending or passive aggressive, which I'm not quite sure what the reason for it is.
Wow, this is embarrassingly riddled with typos - but hopefully my point somewhat came across!
I have some thoughts on this situation as I first saw this on Twitter and now here on lobste.rs the only context we see is this one discussions comment.
I haven't seen many of the other comments made on the nvim-treesitter issue, but I do want to raise a couple of points and my own opinions and I would say context that changes what I personally think about this conclusion.
Additional context which is important in terms of framing this in the context of neovim, Neovim from release-to-release has many breaking changes, the complaint which shushtain had in https://github.com/nvim-treesitter/nvim-treesitter/discussions/8627#discussioncomment-16440673 was that the version 0.12 was not even packaged anywhere and that one of clasom's commits which is just 9 lines of changes in total is the only thing that removes 0.11 support.
I do think that this comment by shushtain is unfairly phrased but I do agree with the sentiment, so I do think that when a maintainer responds in a manner that is antisocial they can be called out on it.
From what I have read from this as a total, is that it is an incredibly frustrating situation definitely for everybody involved. Clasom wanted to rewrite a project with a lot of technical debt in a new branch, however the imperfect way this rewrite was trigged caused a lot of problems and miscommunications, especially how considering many people recommended which I think were fair changes to the README.
And how understandably people can be frustrated that somehow a plugin that used to work, doesn't anymore. And for shushtain I can understand him especially for his comment, requiring nvim 0.12 when that release has been out for literally a single day can be an unreasonable expectation especially considering it's a 9-line commit. that removes the support.
And I really do feel for the maintainer, it is unfortunate to have to maintain literally anything for neovim considering their aversion to anything stable, neovim themselves have breaking changes all the time.
One day I was having a very bad day, and at the end of that day I sat down to unwind with some code writing time, only to discover that an issue I had opened against a popular open source project - and to which I had contributed a fix and was just trying to figure out how to get through their contribution process - was "closed due to inactivity".
Dear reader, I had a bad moment. I said some not very nice snarky things. The lead maintainer came in to the project and - quite rightly - scolded me. Told me that that sort of thing is what causes burnout in open source work.
I was able to get a hold on my emotions and with their help got my fix to land, and I apologized - although maybe not as well as I could have. I think about that experience a lot. It's there in public. I've learnt from it. I hope the people in the discussion (linked elsewhere in comments) that led to this archival have a similar moment of reflection.
Thank you for sharing this. It's very valuable and relieving and hopeful to actually hear that behaving in a somewhat "strong"/"firm" way (but still sincere, showing some kindness, and remembering there's a human being out there) towards a person exhibiting a perceived "trollish" or otherwise excessive behavior can indeed be beneficial for that person, in shorter or longer term. Not that I had that much of it in the past, but I expect I might see more in the future, so I'm grateful to hereby feel reassured in that view, in preparation. So: to remember there's a human, help them learn, but also be firm in kicking them off quickly and just letting go if needed, so that they can have an opportunity to get their actual reflection time and space.
In your defence, stale bots are a blight on bug trackers (they may be affiliate for work trackers, but are strictly bad for bug trackers), and the maintainers are in the wrong for using one. They started the hostile behaviour that causes burnout.
I asked on the Zed discord server why Max or the Zed team aren't active in maintaining the Treesitter project. @MorganZed said:
"I cannot speak for max in his personal capacity but it is not zed nor zed affiliated"
How does this relate to the Tree-sitter project? This (news/post) is a Neovim plugin implementation of Tree-sitter, not Tree-sitter itself.
The audacity of some people. Just pin a specific commit if it breaks and move on, this type of hostility is uncalled for.
I think what's really unfortunate is that people like Clason and SaveTheClockTower shoulder so much of the burden of supporting Tree-sitter while the Zed team largely ignores their own product.
What is the relation between Zed and treesitter?
Max Brunsfeld built tree-sitter and is the plausible authority figure in the ecosystem. He's also one of the cofounders of Zed: he came over from the Atom team.
It's interesting how you are showing same behaviour as users of nvim-treesitter to Clason. Is Max not allowed to be burnt out of the project?
My point was more that his employer is a funded editor company with a fair amount of cash which uses Tree-sitter as a major part of their product. There's any number of ways they could invest in the ecosystem more proactively, and I'd argue that doing so would be in the company's best interests...
I think Sam Altman is onto something here. I know the guy is cunning but I think he has special ability is seeing and grabbing into opportunity. For your average software project, you need a lot of open source and some of it is just critical to get anywhere and a foundational part of this is software tooling.
Contributing to treesitter is not in Zed best interest as it helps free/open editors. It'd make more sense for the experience in nvim to become worse as it draws people who don't have time to monetized products like Zed.
treesitter getting worse directly makes Zed worse by virtue of nearly every supported language in the editor using treesitter. Please stop inventing conspiracy theories.
Isn't this nvim-treesitter and not treesitter though? Treesitter is still active: https://github.com/tree-sitter/tree-sitter/tree/master
your comment never mentions nvim-treesitter, and the parent comment is talking about the main treesitter project not the nvim-treesitter plugin.
Yeah I was at least in part talking about the main tree-sitter project. That is what the Zed team disavowed any role in owning or maintaining when I asked them about it, which surprised me. Max himself has minimal commit activity in the main tree-sitter project since mid-2024.
Actually I think the opposite is true. If Zed wanted to they could be revamping and rewriting tree-sitter at such a pace that no volunteer developer could keep up. It’s in everyone’s best interest if a project is community maintained and not controlled by a single company.
I think my comment was misconstrued. I didn't mean that Zed would actively make open source worse for their own benefit rather, that they have no incentive to improve other open source projects that are already declining due to shifts in the ecosystem.
You are not entitled to the attention of others. You are not entitled to having value attached to your complaints. You are not entitled to this explanation.
If you have expectations (of others) that aren't being met, those expectations are your own responsibility. You are responsible for your own needs. If you want things, make them.
Rich Hickey, open source is not about you
Edit: I assume, like Inicola, that it had to do with the behavior of entitled users. So this is a response to that.
Rich Hickey, open source is not about you
My favourite phrase there:
Morale erosion
When a small landslide occurs, at a cliff's edge, one can not simply push the eroded part back up to the edge and expect it to remain steady for a future trampling.
"Behavior of entitled users" doesn't capture the situation. https://lobste.rs/c/j5rkwx
I applaud the maintainer for doing this. I hope they have a good break, maybe come back, or maybe not. Shame on these entitled users.
I've made a comment about the whole thing, I think blaming it all on entitled users isn't particularly accurate, https://lobste.rs/c/j5rkwx
We should normalise open source projects being yanked as a response to the unsustainable ecosystem.
A bit of a wider interpretation of "as is" where it can also mean "as isn't".