Because Windows malware can't infect Linux, so why would that be a security issue for Linux? But if you run Windows, you definetly don't want to sideload and use unmaintained libraries and APIs that are 20 years out of date.
Because Windows malware can't infect Linux, so why would that be a security issue for Linux? But if you run Windows, you definetly don't want to sideload and use unmaintained libraries and APIs that are 20 years out of date.
To put it simply, it doesn't really matter what libraries you ship, they _cannot_ cause _new_ security issues in the operating system, by simple definition of user space.
And Windows malware can definitely infect Linux. That was not my claim.
Not really, search for UAC virtualization.
> My tax returns, my pictures, my passwords, all of the data I actually care about is stored in files accessible in user space.
So are the passwords of everyone logging in to this very website (stored in user space). I think you are confusing user/privilege separation with kernel-userspace separation.
You have not yet made a point. How can distributing a user-space shared library, no matter how fully loaded with ancient security holes, decrease the amount of security of your system?
We literally have this on Chen's today: https://devblogs.microsoft.com/oldnewthing/20221011-00/ Totally sure malware authors are going to compromise the files of an ancient game in order to trigger some bug in a library to get to your tax returns. No way they will not just change the game exec or something.
They probably want to put pressure on publishers to use newer libraries that don’t need administrative permission, so hopefully eventually getting a version of the program that doesn’t need admin. Encouraging better security hygiene.
I agree that user-space compromise is still really really really bad.