Earlier quoted context omitted.
There’s a moral obligation to mitigate harm caused by your project. I recently ran into a situation where a very old package caused terrible damage. I contacted the pypi maintainer. He apologized and promised to fix it. Six months later, no changes. This was a very unusual situation, as the package was the same name as a module later adopted in the standard library. The author was under the impression the package was…
What was the license of this package?
Atomicwrites' old versions have been purged from PyPI
41–50 of 75 posts
Re: Atomicwrites' old versions have been purged from PyPI
#42Earlier quoted context omitted.
His right
I dunno, I think if you publish a copy of your code to a registry then it would be both desirable and reasonable for that copy to be immutable. Allowing the deletion of published libraries can have huge downstream impacts and ultimately makes the registry less trustworthy. Edit: to be clear, not trying to shame the author here - it sounds like they tried to avoid this situation: "what i didn't consider is that this w…
Life is not immutable. There could be a claim about IP, malware, whatever.
Versions should be immutable, but possible to delete
Re: Atomicwrites' old versions have been purged from PyPI
#43Earlier quoted context omitted.
I dunno, I think if you publish a copy of your code to a registry then it would be both desirable and reasonable for that copy to be immutable. Allowing the deletion of published libraries can have huge downstream impacts and ultimately makes the registry less trustworthy. Edit: to be clear, not trying to shame the author here - it sounds like they tried to avoid this situation: "what i didn't consider is that this w…
I agree. The logic is similar to why you can't delete an HN comment once someone replies.
Re: Atomicwrites' old versions have been purged from PyPI
#44Earlier quoted context omitted.
These weaknesses are implementation specific. FIDO2/U2F is unphishable, requires proof of presence, and is a significant security win over a strong password.
Is PyPI requiring maintainers to use a hardware key? If not, I don’t understand how this policy is helpful. Anyone who hadn’t already turned on 2FA is going to use the most frictionless so-called second factor they can.
Re: Atomicwrites' old versions have been purged from PyPI
#45Earlier quoted context omitted.
Is PyPI requiring maintainers to use a hardware key? If not, I don’t understand how this policy is helpful. Anyone who hadn’t already turned on 2FA is going to use the most frictionless so-called second factor they can.
They've been offering people hardware keys for free. https://pypi.org/security-key-giveaway/
Anyone who is turning on 2FA because of this requirement is going to select the most frictionless method of complying with the mandate. Which will not be a hardware key.
Re: Atomicwrites' old versions have been purged from PyPI
#46Earlier quoted context omitted.
I agree. The logic is similar to why you can't delete an HN comment once someone replies.
Well sure. But what happens if your post contained some confidential data? It gets redacted
Re: Atomicwrites' old versions have been purged from PyPI
#47Earlier quoted context omitted.
They've been offering people hardware keys for free. https://pypi.org/security-key-giveaway/
That's a great initiative, but I expect the maintainers who are interested to be the ones who've already turned on (some lesser form of) 2FA voluntarily. Anyone who is turning on 2FA because of this requirement is going to select the most frictionless method of complying with the mandate. Which will not be a hardware key.
Re: Atomicwrites' old versions have been purged from PyPI
#48Earlier quoted context omitted.
That's a great initiative, but I expect the maintainers who are interested to be the ones who've already turned on (some lesser form of) 2FA voluntarily. Anyone who is turning on 2FA because of this requirement is going to select the most frictionless method of complying with the mandate. Which will not be a hardware key.
OK? So more people use TOTP and there's a marginal security win. And maybe a few use a token, and there's a significant win.
I think the statement "these weaknesses are implementation specific", while true, is irrelevant when 99% of people affected by this mandate (and 99.9% of 2FA users in general) are going to use an implementation with these weaknesses. And, I think it really sucks that PyPI is loosing maintainers due to a policy that won't increase security in a meaningful way.
Re: Atomicwrites' old versions have been purged from PyPI
#49Earlier quoted context omitted.
> PyPI offered to provide a security key to make the maintainer's life easier It's even easier to just leave 2FA disabled and stop maintaining the project. Which is what they did. Are maintainers obligated to support their projects indefinitely?
There’s a moral obligation to mitigate harm caused by your project. I recently ran into a situation where a very old package caused terrible damage. I contacted the pypi maintainer. He apologized and promised to fix it. Six months later, no changes. This was a very unusual situation, as the package was the same name as a module later adopted in the standard library. The author was under the impression the package was…