If it's not fun anymore, you get nothing from maintaining a popular package | Hacker News Reader

If it's not fun anymore, you get nothing from maintaining a popular package

gist.github.com

317 pointsjashkenas133 comments

Personally I felt people on GitHub were being way too harsh. He built something, moved on, and gave it away. He had no way of knowing the person who took it over would use it maliciously.

Seriously, some of the comments blew my mind. Because he built it, he had a responsibility to maintain it forever, and was responsible for everything that happened with that code forever?

Maybe calling free software "Open Source" was a mistake. If something is "free", then you have to accept you may get what you paid for. If you are using somebody else's code, is it their responsibility to look after it years after they have moved on, or your responsibility to make sure it's not full of bugs or malicious code.

Kind of sad that people are more angry at Dominic than the malicious updater.

Seems there is a bit of a "boys will be boys" phenomenon going on.

To an extent it's because his name is in the repo name (dominictarr/event-stream) https://github.com/dominictarr/event-stream so he is an easy and obvious scapegoat. But it seems people in general forgot that they are not paying and are not entitled to any sort of support.

People in this case aren't asking for support, they're asking for the project to have been given to a trusted maintainer or declared abandoned.

To be honest I wouldn't expect an abandoned project to be any different. I would expect it to have lost its maintainer, not switched from a trustworthy maintainer to a malicious one.

Edit: Clarified second sentence.

No maintainer is malicious...until they are. Be honest with yourself: how many dependencies have you vetted the maintainer before including? If n > 0 then you're just as susceptible.

I wouldn't just add some random maintainer with a new GitHub account to a project I controlled. If they had made some contributions via pull requests, I would consider it or if they were a maintainer of other projects with actual users. In this case, there was no vetting at all. A brand new GitHub account with a couple forks.

I'm not placing all the blame on the original maintainer as I understand his philosophy of handing off the torch, but I think it's a naive thing to do, as this has shown.

Ultimately, this has been a good demonstration of a weakness with the ecosystem and I think there are a lot of things that could be improved here to help mitigate this in the future.

How do you suggest finding trusted maintainer who would not turn malicious at later date?

I don’t. No guarantees of anything. But we can probably do better than random Chinese dude with no other projects or contribs

I think people are just drawing a distinction between fault and responsibility. Neither is a subset of the other.

This, I think, is a sucker's argument. Responsibility is not putting trust in a dependency they themselves have no control over. Responsibility is not letting yourself get into an obviously bad situation. It's not even the first time this has happened to NPM. Remember left pad?

No, this thread didn't happen because a bunch of developed chose to make this a conversation about responsibility more than fault. They're mad they got suckered and Dominic is the best scapegoat.

> This, I think, is a sucker's argument. Responsibility is not putting trust in a dependency they themselves have no control over.

Responsibility can fall on multiple people. I don't understand why everyone looks at things in such a binary way. Responsibility, fault, etc. are not binary things. Two or more people responsible for one problem can be both upset at each other perfectly justifiably.

Responsibility is a hard problem. It's quite true that more than one person can be responsible for something, in part, but there's only so much responsibility to go around: whatever damage irresponsibility causes, more than one person cannot be responsible for the full damage, if we're going to maintain any consistency.

Authority and Responsibility go hand-in-hand. If you have authority over something/someone, then you are responsible for it/them. You cannot be responsible for something you have no authority over, and you should not have authority for something you have no responsibility for.

In this case, every developer has authority over their own code base. If I decide to include some random code from the internet in my project then I am responsible for that decision. If whoever has commit privileges on that random code decides to commit malicious code, then they have responsibility for that. However, that doesn't make them responsible for my project, because they have no authority over my project.

Including a dependency in a project is a security vulnerability. It may be worth the risk, it may not, but it is definitely a risk. And responsibility for taking that risk is entirely on the person who took the decision to include that dependency.

> whatever damage irresponsibility causes, more than one person cannot be responsible for the full damage, if we're going to maintain any consistency.

This may be a moot point if you consider that "full damage" can be more than what one person can be expected to take responsibility for and handle.

The entitlement mentality of some people is insane.

Especially when they aren't paying.

Sayre's law, baby.

Intensity of feeling is inversely proportional to issue value sometimes. Recent US political discourse seems to violate Sayre's law.

>He had no way of knowing the person who took it over would use it maliciously.

Then it was, clearly, negligent of him to give it to this person. He gave a stranger the power to do things in his name.

He doesn't have to maintain the project forever - it seems like a much easier solution is: don't give away a project with your name on it to someone you don't know. Instead, encourage them to fork and release a new version under their own name. This is a surefire way to never have an exploit spread to thousands of people with your name on it.

As a user of free software you should always accept the risk that it is maintained by a malicious actor. It is just as likely that the original author will turn malicious. You as a user have the responsibility of auditing all the free software you use.

People keep casually making this suggestion. In JS this can involve hundreds or thousands of transient dependencies. Have you ever (1) worked on a React app and (2) vetted all of React's 829 dependencies yourself? It's not remotely feasible. The only solution becomes "then don't use React!" - is this the best we can do?

React does not have hundreds of dependencies:

  └─┬ react@16.6.3
    ├─┬ loose-envify@1.4.0
    │ └── js-tokens@4.0.0
    ├── object-assign@4.1.1
    ├─┬ prop-types@15.6.2
    │ ├── loose-envify@1.4.0 deduped
    │ └── object-assign@4.1.1 deduped
    └─┬ scheduler@0.11.2
      ├── loose-envify@1.4.0 deduped
      └── object-assign@4.1.1 deduped

React Native then. 1200+ last I remember.

The point, I think, is that this is the real problem, and while Dominic might have made it harder for the attacker to get this exploit into the wild, there is nothing that he could have done that would have prevented such an outcome.

If one is writing something as security-critical as a Bitcoin wallet, then not using React (and other things with a similarly wide attack surface) is indeed the best you can do. Security is not tolerant of half-assed measures.

It is much less likely the original maintainer will turn malicious, especially a trusted member of the community.

> It is just as likely that the original author will turn malicious

it absolutely isn't. The original author is a real person, with a twitter account, with a profile photo, with videos of talking at conferences. I accepted the code in my project because such author wouldn't get away with injecting the code with a backdoor. But a random alias that quietly takes over a package I didn't even know was abandoned because the author was too lazy to write "This package is no longer maintained" in the readme? That didn't give me any chance...

Depending on the author's reputation isn't much better. Maybe the author uses a weak password and doesn't have 2FA on his npm account. You have absolutely no idea.

The latter solution isn't much better - all the community momentum of the first project is instantly gone. Now you have a set of forks, all with a different small set of patches on top. Which are maintained? Which will be maintained going forwards? Any new user looking for this functionality has no idea. Maybe none of them.

He didn't give anybody the power to do things in his name. He gave a stranger the power to contribute to a shared codebase. He never claimed to be a BDFL of this codebase. Your vitriol would be much better aimed at node modules which depended on non-fixed code which they didn't audit.

losing the "momentum" on an abandoned package is a much better choice than having it fall into the wrong hands together with all the people who use it and didn't even know about the change of owner.

That's a false dichotomy. The actual options are:

1. Dead project

2. You share access to lots of people and they act responsibly

3. You share access to lots of people and they act irresponsibly

#2 is waaaay more likely than #3.

Still seems to me like people are blaming the wrong person here. If you're writing a cryptocurrency-related module (the target of this specific attack) and you're using npm modules written by somebody you don't know, what the hell are you doing?

#2 is more likely than #3 because a reasonable person gives access to another contributor or someone they spent a few minutes vetting, not a total stranger with no GitHub history.

Exactly. Both ways have have costs and benefits. It's certainly easy take this one example and say, "All projects should optimize for this very rare failure case." But you can bet that if he just closed his 700 packages down, we'd have plenty of people explaining how that was the wrong solution, that he should work to find maintainers.

As far as I'm concerned all of this "he should have" stuff is more proof that he's right: maintaining a popular open-source package for free can be no fun at all.

>The latter solution isn't much better

It is unquestionably better than your users, whose trust you've earned over time, suddenly finding that their app is mining cryptocurrency.

>He didn't give anybody the power to do things in his name.

His name is still on the project he gave away access to, so yes, he did.

>Your vitriol would be much better aimed at...

I think criticism is warranted here, but I don't think I've been especially vitriolic about it. Frankly, OP made a poor choice and is now making excuses by complaining about not being rewarded for maintaining the project rather than own up to it. There was a better way to do this without him having to perpetually maintain the project.

And moreover, the herd of commenters here on HN are projecting their dissatisfaction with entitled open source users onto anyone who criticizes OP.

Users have to come to grips with the possibility that the "free" code they download might be malicious.

To mitigate the risk they have these options:

1 - Don't use the software.

2 - Audit the software themselves before they use it.

3 - Pay someone to do the auditing for them.

Even if they choose options 2 or 3, there's still a chance the some vulnerability of malware could sneak through (especially if they're getting a binary blob or source code written in a language like C that makes obfuscation easy).

But if they choose to use unaudited software, they're playing with fire and shouldn't be surprised if from time to time they get burnt.

I wonder if there are projects out there for an option 4:

4) Use an OSS dependency checker tool that reports the risk in your dependency tree

4a) Dependency tool uses a crowdsourced database of audit/review metrics

A typical NodeJS project has potentially 1,000's of (transitive) dependencies. Many well-established projects use dependencies versioned e.g. 0.0.1 etc.

I guess there are hundreds of similarly unmaintained packages among these.

Lacking 4) a small-time developer working on a side-project only has option 1) (or option 0 - Take the risk) as a feasible way to go.

Users have to come to grips with the possibility that the "free" code they download might be malicious.

denial, anger, bargaining, depression and acceptance.

     ^^^^^

People are dealing with a lot of negativity in their life, they shouldn't have to deal with that entitled bull-crap online, especially when doing unpaid work. Should the maintainer have been more cautious? no, he should have straight out closed the project and just ask people to fork it or use something else. If you are not going to maintain something anymore, don't even bother worrying about passing the torch. That way nothing is your responsibility anymore. The people who complain, insult the maintainer in Github issues right now will be at the root of a shift of mindset from contributors that will end the "golden age of open source".

I think this is the clearest, easiest, correct answer. If you no longer have the time and motivation to maintain a project, and are also not in a position to hand over control to someone you have good reason to trust, the easy, safe answer is to retire it, and have others fork.

Then in the original project you can say “the last version of this project is N.N; here are some possible candidate successor forks, which have not been vetted”.

This way no one who is merely keeping up with version numbers will accidentally step off of a reasonably good package on to a fraudulent package.

>I think this is the clearest, easiest, correct answer. If you no longer have the time and motivation to maintain a project, and are also not in a position to hand over control to someone you have good reason to trust, the easy, safe answer is to retire it, and have others fork.

To prescribe some preventative advice based on this: it's probably not a good idea if your project gets popular to continue being a solo developer in open source. So plan ahead based on your popularity.

And your project isn't likely to explode in popularity over-night and be a dependency for 1000+ other projects, so you should be able to see this coming.

Of course, this is super-easy to prescribe in hindsight.

I 100% guarantee you that if he did that, he'd be getting a lot of static about how he should have handed the package over to some contributor.

Which other similar situations are you thinking of?

left-pad

left-pad was a very different situation. The maintainer unpublished the module from npm. I forget the reason, but I want to say it was some sort of rage quit.

he had a package called kik, which npm dropped for the benefit of kik the company (though I don't know whether they requested they do so). In response the author unpublished everything from npm.

“You can please some of the people all of the time, you can please all of the people some of the time, but you can’t please all of the people all of the time”

This is mitigated if enough notice can be given in advance.

I agree that's what he should have done. Isn't that being more cautious?

I'm being naive probably, but I have a hard time believing that this controversy would be so big had it been a case of an abandoned library. Users do whine a bit in the Github Issues for dead projects, but at least they can tell the project's dead and/or broken. In this case, the maintainer, who is a security professional, made the active decision to give ownership to an untrusted party. He shouldn't be held liable in any legal/financial sense, but why wouldn't users be alarmed/unhappy?

> People are dealing with a lot of negativity in their life, they shouldn't have to deal with that entitled bull-crap online, especially when doing unpaid work.

They are practicing resume driven development. While this is unpaid, it's not completely altruistic.

It's impractical, on an individual level, to vet your entire dependency tree. Installing something like react alone will put 100+ modules in your dependency tree.

A more mature project will have a 4-figure number of dependencies.

Things like Node Security Platform (which was absorbed by NPM) exist for a reason. There's no reason for every person to vet every single package they use -- it'd be an awful lot of duplicated energy.

If there was no implicit trust in the community, and we had to act as vigilant as you describe, there would be no community.

This is a bit of a strawman. It absolutely is your responsibility, but that doesn't mean you can't outsource it.

It's the kind of thing where, if everyone with this problem chipped in $5/month, we could have a central repository of vulnerabilities. So, not absolving yourself of responsibility, but also way more efficient to outsource.

(React does not have 100 dependencies, it has 5 including indirect dependencies)

A more mature project wouldn't have just React as a dependency. Point is, it's a lot of code to vet. React may have 5 indirect dependencies, but in my projects, I have 30 or more direct dependencies.

I experienced a twist on this. A guy who had very different aesthetic taste wanted me to merge some patches. He got real entitled about it, eventually I banned him. THEN he did the "appeal to Twitter" and made malicious edits to the project page on wikipedia. He made a huge stink and wasted hours of both of our time. It was unreal! I wish I was joking.

I wrote about it, discussing my motivations for open source:

http://charlesleifer.com/blog/your-idea-sucks/

A. Damn, I would think a CTO would know to quit this dispute and fork his own purportedly superior version, if his company's survival is so dependent on it.

B. Really like the elegant style (visually, though the writing is also elegant!) of your blog.

You shouldn't have hidden his name. We should know who it is, in case we meet him or something.

Then I'd be doing exactly what he did, just dressed up a little in a blog instead of twitter.

No. Complaining about horrible behavior is not horrible behavior about itself.

I caught that notion they had

> I've shared publish rights with other people before. Of course, If I had realized they had a malicious intent I wouldn't have, but at the time it looked like someone who was actually trying to help me.

Can anyone cite perhaps a handful of examples where this malicious open source usage ultimately bit them?

Well they certainly couldn't call it "free software", that involves many more assumptions than "Open Source", and it might not even be free of charge. :) Blame Richard Stallman for using an incredibly common word with many meanings to refer to something that has one incredibly specific meaning which is unintuitive to most people, and to pretty much anyone who isn't a developer.

I think several decades of Open Source boosterism may have eventually given some people the wrong idea about what it's all about. Two annoying trends I've noticed in the last 10 years:

(1) (what this poor guy seems to be running into) People treating Open Source as anything other than charity from the people who put it out there. The people who make it generally don't owe anyone anything, including the people who use their code. I sometimes see very harsh criticism from people within the community that assumes anyone putting anything out there has to accept all sorts of responsibilities: maintaining a community, keeping on top of bug reports, acting as an integrator for people who want to send in patches, making sure that all packaging and documentation conform to some very high bar and/or specification, etc. Just because somebody puts something out there doesn't mean it has to be a second job for them, and doesn't mean that they are obligated to do anything for anyone, including responding to emails or dealing with bug reports or patches. It's perfectly fine to stop maintaining something after a while, or to just throw code over the wall and never touch it again after it's out there.

(2) (not this case) People treating Open Source as some avenue to professional fame or an eventual source of income. Yeah, that can happen, but it very rarely happens, and it's not a good reason to write it in the first place. This is the result of too many people using business arguments as a fundamental justification for doing Open Source, and people saying stuff like they only hire people who have projects on Github, and only if they like what they see. If someone's primary motivation is fame or money, Open Source is a terrible way to accomplish that. It's possible, just highly unlikely, plus I think it's much less likely to happen if that's someone's goal from the outset.

I think there is an underlying root cause to how we treat Open Source, and it's similar to the IT "cost center vs value creator" problem: Open source packages are treated as consumer goods, and creators in turn will engage with that by marketing them as consumable products.

Many of these things which are distributed as packages are actually code snippets. They could be maintained in the form of a Stack Overflow answer to copy-paste into your code. The question-answer system exposes the code to the necessary competitive evaluation: a malicious answer soon gets downvoted. When the packaging metaphor is laden over them, the whole chain of responsibility lapses back to the vendor-consumer model. And this works when packages are large and take pains to reduce their own dependencies, even if it means common functions get duplicated across different packages.

But prolific authors currently are rewarded by being vendors of undersized packages: they enjoy large download numbers, Github stars, and so on. Without any kind of curation to impose friction on this process, they can do the wrong thing for a very long period and get zero feedback indicating that their actions create a problem. And on the consuming side nobody pays a lick of attention when the prompt says it's going to install 50 things with cute-but-opaque names. If they are under a page of code each, you could probably just print the code.

When I create projects now, I tend to grab some packaged code to start with, but apart from a few to do core functions like I/O I will treat them as placeholders and seek ways of reworking the code to make myself the sole maintainer. The alternative is to be a software consumer and to complain like a consumer - i.e. in a largely powerless fashion - when other people's code doesn't do what I want.

> Maybe calling free software "Open Source" was a mistake.

This conflation is definitely a bad thing. Words with an actual meaning get over-used, simplified, stripped of their original meaning, and then merged. It would be 100% to the detriment of our language if it happens in this case.

Open source does not mean free (as in freedom or beer), free (as in beer) does not mean open source. Free as in freedom almost always means open source.

Side rant: Also, open source doesn't mean "open to contributions". Giving access to the source code does not mean obligation to accept contributions to said source.

> Side rant: Also, open source doesn't mean "open to contributions". Giving access to the source code does not mean obligation to accept contributions to said source.

How true, this is such a common misunderstanding among newcomers to FOSS, it can't be repeated enough times.

This problem can be solvable by introducing “Verified Identity” on GitHub. Anyone who wants to be maintainer of popular project can be asked to get Verified Identity. If this person turned out to be malicious, he risks getting tracked down by law enforcement fairly easily. For getting Verified Identity, simplest thing might be validation of phone number. Bit more complex thing might be Verified bank account or credit card. Assuming GitHub has 1M active users and each user verification costs $5, the cost of solving this problem is less than 0.1% of GitHub valuation but can hugely improve trust levels.

Good idea although $5 to a person writing malicious code and out of jurisdiction isn't much of a deterrent IMO.

Why on earth would anyone want to be on the hook for liability like this for an unpaid maintenance job?

Open source licenses state this clearly: "THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND"

Agree with you, just note that the term "free software" doesn't actually refer to software price at all.

> Maybe calling free software "Open Source" was a mistake. If something is "free", then you have to accept you may get what you paid for.

Since no one seems to care about the freedom aspect of it, maybe we could rename it to "fee software" - give it away for free and encourage everyone to contribute with time and money. Will people stop being selfish because we changed the name? I don't think so.

The community discussion is focused on the wrong thing. Here is the problem:

https://github.com/dominictarr/event-stream

Look at that URL. It has "dominictarr" in it. Now click on the URL. Look at the big "dominictarr/event-stream" in the upper left of the screen.

Many are going to conclude that this package belongs to dominictarr. Almost everything in the UI indicates this.

Developers don't trust code, they trust people and organizations. NPM and Github needs to make it very clear when ownership has changed. And needs defaults that properly inform all users when ownership has changed. For example, when I SSH into a server the first time, it asks if I trust the key. Then one day, what if it asks again? Then I can stop and investigate why the key has changed, what's going on, and if I should trust it again.

With the current NPM/Github there's no warning to users. A developer could've checked dominictarr's background and decide they are trustworthy, which would then extend to event-stream as well. But where's the warning to users when ownership of event-stream has changed and they need to re-evaluate whether to trust the new owner? And why does it look like the project still belongs to dominictarr when it actually doesn't? Those are the key issues here.

> Developers don't trust code, they trust people

well the code doesn't write itself, so partially true. But ultimately it's about accountability. If Dominic himself added the backdoor, his presence in the community would end that day. If an anonymous alias does that, we don't even know who to blame. Dominic's fault is not the backdoor, but the irresponsibility of handing over a used package to an anonymous alias. Not being clear about the fact that the package is abandoned. I wouldn't depend on an abandoned package, if I knew that.

> Not being clear about the fact that the package is abandoned

This is really the key point here.

The NPM organization could go much further to make these attacks harder.

You pitch a really good one: Any time you npm update/install, display ownership changes (especially compared to your prev version).

Another one is to show the source code on the NPM website itself instead of hiding it in a tarball. NPM basically trains people to assume the published code == the code at the linked repository. It's a hacky honor system that only helps attackers.

Another easy suggestion is that NPM could have forced a semver major change on the new maintainer. It would have been an easy signal for people to check what changed, and fewer developers would have accidentally installed the infected version because it was only a "minor" change.

I feel bad for the guy, he's not wrong, that's for sure, but a recent example of "doing this right" is here: https://news.ycombinator.com/item?id=18525025 / https://github.com/FineUploader/fine-uploader/issues/2073 (FineUploader is shutting down)

At least, it seems that way to me. If I were ever in a burdensome positioning of maintaining a package I didn't want to maintain, I'd go for the explicit archive/sunset route. At that point if someone wants to fork and re-package, they can do it, but it won't mess with anything directly downstream immediately.

I agree that the author has every right to quit being a maintainer, for any reason, but I don't believe that he is necessarily absolved of all judgment just because it was a free project he did on his own time.

He's obviously a prolific and generous author, but that experience also means he's very aware about real-world practices (especially in the npm ecosystem) to know that ceding control of a popular library, without taking any precaution, has the potential to put a significant number of users at risk. Should the users be less complacent? Sure, but that's not currently the world we live in, and the author is not ignorant of that.

If an experienced electrician volunteers his time to a homeless shelter, and decides to quit halfway through a rewiring project, it's his right to quit whenever. But the manner in which he leaves the project is also important. There's a huge difference between:

- Confronting the shelter manager and saying "Hey, I can't do this anymore, and you get a professional ASAP to fix my work because I can't guarantee that I've properly grounded everything"

- Exchanging pleasant goodbyes with the shelter staff, and not giving them any reason to worry about what you may or may not have done.

The latter situation is akin to a silent failure -- even stomping off the job and telling the shelter to go fuck themselves is ultimately preferable, since the shelter staff at least are very aware that your work may be dangerously unfinished. While you may not be liable if anyone gets shocked/electrocuted from your leftover work, that doesn't mean the shelter staff are obligated to not grumble and warn anyone else who might be relying on you for volunteer work.

edit: that said, I don't have a major objection to the author's response, as submitted here. I was referring to people who think that any significant criticism is unwarranted, just because the work is free. The author, to his credit, made a decently detailed reply and continues to make thoughtful responses.

I think your example makes your point weaker rather than stronger, because unlike the electrician, this guy has no reason to believe he's done a bad by job himself, because he's not putting anyone's life in danger, and because others who are using his work are also supposed to be similar professionals in their technical fields and taking similar responsibility. Comparing it to laymen who know potentially nothing about secure wiring getting potentially electrocuted because of a malpracticing electrician is not a great comparison.

> and because others who are using his work are also supposed to be similar professionals

I have a hard time believing that anyone who has built so many popular packages on npm (or any open source) also believes that the hundreds of thousands of users who install his work are "similar professionals" to him, or, even if they were, that they have built safeguards against this kind of scenario. I've barely used npm/node in the past few years and even I'm still aware of how the ecosystem makes silent changes to dependencies non-trivial to detect.

> because he's not putting anyone's life in danger

Software has real-world implications.

I'm trying to help you make a potentially good argument stronger, not trying to shut it down. Disputing it and telling me I'm wrong (even if I am) and that electrocuting people is like JS frameworks having "real-world implications" is not going to help you with that.

Thanks for the help, but I don't see the need in perfecting the analogy. My argument is that volunteer work can be criticized, despite its altruistic intentions. I have no interest in debating with you about how software impacts lives. Apparently, this case has enough potential impact to be a much-discussed controversy.

To use your scenario, but slightly change it:

- Tell the manager you will no longer be doing any work.

- Not tell the manager you’re quitting and replacing yourself with a random stranger from Home Depot who seems interested.

It’s not wrong for the electrician to quit, especially when volunteering, but it’s wrong not to properly notify people and to just let a random stranger take your place without notice.

So much for "many eyes make all bugs shallow" I guess. Problem is that 99% of those eyes are turned only toward using the product of someone else's good will for their own personal profit. And 99% of the remainder are only looking for bugs in the features they use, or just aren't that good at spotting bugs. This only gets worse in environments like NPM, which make it easy for dependencies to proliferate beyond anyone's expectations, so people don't even know what code they should be looking at.

This is still better than the situation with proprietary software, but problems like these are kind of inevitable in an ecosystem where people typically write 5% of the code themselves and blindly pull in the other 95% from who knows where via unchecked package dependencies.

Those "many eyes" are the reason this was caught at all. I could easily imagine situations where this backdoor would go unnoticed for months or years.

Some tiny subset of those many eyes might have found it, but by then its impact was far from shallow. In some open-source ecosystems (e.g. GNU or BSD) the ratio of useful eyeballs to vulnerable ones is much higher. It almost seems to be a goal of NPM to maximize the blast radius before a problem is noticed. That's not a healthy way to manage open-source projects.

I've written a lot of software in my life -- almost all of it for myself, and I just never bothered to open source most of it, because if I'm going to open source it I want to make a good impression, so I want to clean up the code and write some docs and make sure it it builds ok for others, think of a name for it, upload it somewhere, etc. It's just so much easier to keep it for myself and not bother with any of that.

Despite all this, I have open sourced a couple of things here and there. Little things. Things that virtually no one ever used, as far as I know. But once I did get a very polite feature request from someone on something that I'd moved on from, so I told them I wouldn't do it.

That was more than a decade ago, and I still think about it and feel kind of guilty for not doing it.. even though I know I have nothing to feel guilty about.

I can imagine being the author or maintainer of a popular package must be like this times a thousand. You have to put up with so much negativity and pressure from people who demand you do work for them for free.

I feel for these people. Anyone who spends even the smallest fraction of their life to give a gift to the world does not deserve anything but praise from those who benefit from it.

> Despite all this, I have open sourced a couple of things here and there. Little things. Things that virtually no one ever used, as far as I know.

You wouldn't know, when learning every help and every practical example can be useful. I understand that not everyone likes to learn by reading someones code but it can be invaluable especially for students in places where books are prohibitively expensive compared to the living standards. You're doing good deed, even if it's just small things, but make sure to document it properly.

Anyone blaming Dominic for not being more responsible should blame themselves for the same reason. This isn't the first time Node has had horrible dependency issues. Have we forgotten left pad? https://news.ycombinator.com/item?id=11340510

Ultimately, you can't take dependencies for granted.

>Ultimately, you can't take dependencies for granted.

The problem is that JS dependencies are so modular that shifting the burden to vet your dependencies to developers is generally not realistic. Creating a new project with one of the current popular frameowrks will bring in hundreds or thousands of dependencies. Who can possibly vet that?

To compound things, this package could be a transient dependency of a transient dependency, so I may not even know the person who decided to depend on it in the first place, or the reason why, or what the package does.

It's not that people are taking dependencies for granted - it's that there is no reasonable alternative.

Dependencies should be nailed to versions and upgraded conservatively. The default shouldn't be the latest and greatest, it should be the last thing that worked.

I agree, but you have to start from a known working state at some point - and at that point you are extending trust to that specific set of versions. Anyone who ran `framework create-project` for a framework that included event-stream as a dependency will have downloaded the exploit and then pinned it without knowing.

Stack (https://docs.haskellstack.org/en/stable/README/) for Haskell solves this problem by maintaining sets of curated version numbers that are known to work together.

This also has security implications as in this approach you don't get the fixes to known vulnerabilities. If you update to 1.2.4 as soon as its out, you may be vulnerable to a takeover like this (which happen but are rare), but if you're still running 1.2.3 when 1.2.4 is out, you're definitely vulnerable to all the things that 1.2.4 fixed, and these risks are far more common.

If semantic versioning always behaved as it should be, the default shouldn't be the last thing that worked but rather the major/minor version of the last thing that worked followed by the latest and greatest patch version.

>If semantic versioning always behaved as it should be

The exploit in question specifically relied on this expectation, by creating a new patch release for the exploit. Even if you could trust well-intentioned maintainers to use it correctly, there's always this risk.

In theory perhaps, but in practice what actually happens is that your dependencies gradually fall so far out of date that after a couple of years it becomes impossible to upgrade anything at all. Then it becomes impossible to add features and before you know it the whole application is dead end legacy code.

There is a reasonable alternative, though...do not use that dependency! It's sure attractive (and easy) to use that one library with a thousand child dependencies, but that's a HUGE red flag that it's not as useful as you think it is.

Incidentally this can come into play even for exceedingly popular packages. For example, consider deciding to use Babel versus TypeScript for ESn -> ES5. I am a fan of both, for different reasons and uses. Both are excellent tools, widely used, of great quality.

Babel, configured in a typical way, will have at least hundreds of (transitive) dependencies. TypeScript has none, every line of code is from Microsoft.

I have pointed this out to numerous people in the process of deciding between these tools; so far I don't think a single one has given this consideration any weight in the selection.

I give it a lot of weight. Microsofts releases are backwards compatible in many cases, and when not, they always have a detailed transition path.

Their changes are never motivated by having to change to suit an upstream packages new version, and their entire codebase is cohesive.

Its what I like in a compiler: a stable tool that I never worry about.

>Creating a new project with one of the current popular frameowrks will bring in hundreds or thousands of dependencies. Who can possibly vet that?

People who care if their software works correctly. Everyone else relies only on herd immunity. If you have dependencies, you either belong to the first group or the second, and you make that choice. Neither choice is wrong, but it is still a choice you have to make and account for.

Does anyone have a big list of issues like this? Both for NPM and for other package repositories. I remember there has been more cases than leftpad. I'm trying to teach some people to not rely too much on them.

I have observed an ongoing pattern. People who create a thing (that becomes somewhat popular) in some sense get credit for it. They are mentioned as the creator of it, forever. It becomes part of their resume, their online reputation. Perhaps they go on to create another thing.

This happens to a far lesser extent for people who adopt a project later to become an ongoing maintainer.

A project might have an originator who works hard for six months, and rewarded forever, and some maintainers who work hard for six years and gets very little reward.

Out of respect for many people who do great work, I will not name names. :-)

Personal attacks are completely unwarranted. I think a more useful (and certainly more constructive) way to look at this is to frame it as a learning experience and figure out what we can do to minimize the effect of this in future.

It's almost impossible to vet every sub-sub dependency of a large codebase without dedicating huge resources. Perhaps there needs to some sort of a javascript code signing standard that requires a level of user identity verification. This is a solved problem with Authenticode and Gatekeeper for executables on Windows/Mac. We have been able to hide behind the fact that we're devs (we know what we're doing!!), but many consumer-level applications (mainly Electron-based) now run node modules with user-level (or higher) permissions. Maybe we need to become a bit more diligent in this area and look at less ad-hoc ways of managing dependencies and identity verification.

Dominic is a prolific contributor and compassionate leader in the JS and P2P communities - and it was a disheartening thread to read. The suggestions that he (working for free!) should have done more seemed unfair.

We should look to fix these:

- If we're going to rely on hundreds of packages per project, we need fine grained security like Android or iOS. deno seems to be doing node right - https://github.com/denoland/deno/issues/171

- Lock down versions for now. Use 1.2.1 instead of ~1.2.1 or ^1.2.1.

- If npm must remain at the center of node's module system, they might have to add auditing and approval tools. 'npm install' can warn about unaudited modules. The problem with that is that it will make npm (a business) even more central to the node ecosystem - which may not be a good thing.

I think this is a major loss of innocence moment for open source. The only sustainable solution to this sort of problem is to pay people to do at least minimal curation and maintenance of open source libraries.

In an ideal world, the payment would go to people who actually create open source value. However, I'm cynical enough about human nature and business to suspect that it will mostly go to middlemen like record distributors and academic publishers, who create very little actual value but somehow make themselves indispensable in the ecosystem.

I found this comment insightful: https://gist.github.com/dominictarr/9fd9c1024c94592bc7268d36...

Summary:

1. Mark as deprecated

2. Change the name by adding "-archived"

3. Turn off Github issues

4. Give away ownership if there is at least a mini-community

but noo, uh .. my fame!!! :'( cries in npm

Is there even a generalized guideline on how to handover npm packages?

It would seem like common sense to let something die and let any new interested maintainers fork the project thank to just wantonly hand it over to a complete stranger. But I could easily see someone getting into the habit of trusting people and not thinking about it too much.

Maybe a combination of better guidelines, some kind of code of ethics along with better means of communicating these changes through npm would've helped this issue a lot more. If not to actually prevent the problem then to at least have some ground to hold people accountable other than accusations of people should know better.

Perhaps you should suggest these guidelines:

https://npm.community/

I don't know what it is about the current environment but it pains me to see people get the life sucked out of them for sharing code and making the world a better place. It's also equally disheartening to see prominent open source contributors essentially close their social media because they're constantly hounded by entitled "community" members and trolls.

> There are too many NPM packages, and worse, too many people quick to add as many dependencies as possible, resulting in crazy dependency trees that you should never actually see in the wild. Additionally, it seems to only be in the node ecosystem that you see single authors with hundreds of packages. Why they do this, I do not know...

Because javascript has a weak standard library.

Maybe the most popular small libraries should be merged together into a single library that can be used to reduce the number of dependencies, in the hope it can become a standard. Like a jQuery for npm.

I think that was the motivation of https://js.foundation

Nah it's less about the standard library and more about the fact that Javascript doesn't have dead code removal so keeping a package small is one way to ensure you aren't pulling in code you aren't going to use.

Of course the failure mode here is that you if you still choose a batteries included framework. (i.e. Angular, or a React powered framework) you end up pulling in a whole bunch of code you don't use and the list of dependencies is unmanageable. But because javascript was born in the browser and most javascript still has to be shipped to a browser it has constraints most other languages do not.

And this is why professionals didn’t depend on build-time fetch of dependencies.

Always makes me think of Brian Fox's canned response to people who flamed him about bash: "Please return your copy of bash for a full refund."

I read this: https://github.com/dominictarr/event-stream/issues/116#issue..., which might seem controversial, but what was he supposed to do?

I guess just let it rot and tell people to fork.

"Since the early days of node/npm, sharing commit access/publish rights, with other contributors was a widespread community practice. https://felixge.de/2013/03/11/the-pull-request-hack.html open source is driven by sharing! It's great!"

Yes, but, quoting from his post he cites:

"So should you really do this for all pull requests? Probably not. While I've given a large amount of users access to various projects of mine, I'm still looking for:

"Github profile: Does this user stand to lose a reputation by doing something stupid?

"Skill: Based on the patch, do I think the user could be a good developer?

"Usefulness: Is this patch solving a valid problem?"

So NPM users et. al. are a hoard of unwashed children. This is like Lord of the Flies.

> If I had realized they had a malicious intent I wouldn't have

Welcome to the Internet

There are people buying chrome extensions, so they could inject something else this guy is surprised when someone did it to his project, which he handed over for free

However you frame it, a very widely used package management system based essentially on trust and a community that doesn't expect any amount of responsibility with regards to those packages seems like a bad combination.

I dont get why the original author is getting so much flack. This is a problem with semver and npm. I hope versions are immutable on npm.

You could argue he could have done a better job in communicating he and his reputation is no longer what is behind this project. But that it has moved to someone else.

I have a rule I don't maintain software I don't personally use. It's kept me from picking up projects other people ask me to do for them, and it's been a good dividing line when it's time to hang it up. I think his major mistake wasn't advertising he'd moved on, and I feel badly for the heat he's taking.

This references https://news.ycombinator.com/item?id=18534392

It looks like it's title has been changed from when I first read it.

Yep. I submitted it with "Dominic Tarr’s Statement on the hacker hijacking of the event-stream NPM module", which was as descriptive of a title as I could think of quickly.

It looks like the moderators have dinged the submission and changed it to: "If it's not fun anymore, you get nothing from maintaining a popular package" ... which I think we can all agree is generic, and fails to make the connection to the big news of the day.

It may just be whispering into the wind, but if the mods are listening — how about "Dominic Tarr: statement on event-stream compromise", which uses the Gist's description, and helps folks know what this link is all about?

That was incredibly well written and I believe him truly. I had not considered that the amount of these tiny packages was probably people trying to learn.

Why wasn't NPM monitoring for this kind of behavior? I guess it wouldn't catch all bad behaviors but this is just asking for it.

What happened to tidelift? https://tidelift.com/

This is a little orthogonal, but I'd really appreciate thoughts on this:

Is FOSS a good idea?

I've always assumed yes. Think of all the libraries in the Python ecosystem, for example. I've used them to build tons of things, even a couple of things that made a little money and allowed me to get myself a new laptop. I've written open source projects myself, filed issues and helped fix bugs in others. The code I've written represents a lot of time, and I'm happy to give it away.

Think about Linux, and what it has done for the world. I've used it for non-commercial purposes all over the place, and it's made research and teaching I've been involved in much easier; it may have been impossible to achieve many things without it.

But recently I've begun to question the idea of FOSS, sparked by Jaron Lanier's concerns. There's something clean about paying someone for their software: you have an entity responsible for the code, you can pay for support, you have recourse if you need it. There are clear incentives and rewards, and money earned is properly redistributed. I know that companies like Google do a lot for FOSS, but it's probable that their contributions are minuscule compared to the monetary value they have extracted from it.

The alternative is not closed-source: e.g. with web dev you can provide the source code but charge for licensing, etc. So it's more about "free" vs "non-free" than "open" vs "closed".

The idea of people coding for free is quite an odd one. Imagine if the world's graphic designers en masse started providing millions of free designs covering every possible icon and situation? (I'm aware of small examples, but they're not at the scale of OSS). Designers don't do it. You would make a lot of people redundant overnight - not all designers, but a lot of them.

What do people think? I can't see how we could have this amazing ecosystem without FOSS... but maybe some kind of transactional funding model is needed to make it sustainable, more secure, etc.

My instincts were for FOSS, but there's often a difference between what sounds beautiful conceptually and the problems of how it works in practice.

RMS' ideas were very attractive, but then you see the modern world from a musician's point of view and it sucks - the long-term effects of ignoring copyright have been to diminish income for musicians. Whilst I improve my friend's quality of life by giving them a copy of some music, their kids might grow up in a world with less good music as a result. I know the record industry sucked, but free copying is not the only alternative.

I'd really appreciate input here, as I have failed to resolved this either way. Thoughts?

A free gift like fire to mankind can be both great if useful and bad if misused. Just nature. And the being helped it forever have an eagle chasing his liver ... or he has to do the stone.

Makes you wonder if we are at Peak Open Source

The largest pure-play open source company and the largest open source host both being acquired within the last 6 months is a strong indication.

the quid is if the maintainer is a douchbag or not

So much good software out there is forever hidden from the world because the creators are afraid to be seen failing publicly, because of shit like this.

To the brave souls who release open source packages, here’s to you.

Reply on news.ycombinator.com