I've been wary of Gatekeeper and SIP as moving Macs towards an iOS-style walled garden, but this is a perfect case of SIP protecting the user from bad software.
More seriously, SIP messages do show up in the system logs, which next to no one ever reads unless it's to find out that SIP is preventing something that the user really intended to occur.
It seems even more likely that the result of unlink() would be ignored (right up there with ignoring printf()), not because it’s the right thing to do but because lazy programmers will assume that failures are incredibly unlikely or unimportant. For example, if the code is a cleanup phase that just wants to remove a list of files, what are the odds that the program dutifully checks that the files actually went away?
Or, as the reason for the omission of such checks is more likely to be, what to do if something that shouldn't fail, fails? And if whatever you decide to do to handle the error itself also fails? Repeat ad infinitum. To even try to go down that rabbithole is simply a waste of effort and does nothing but introduce unnecessary complexity, to put it bluntly.
As a consumer? No. I'm not interested in QAing buggy software. I want it to not break my machine.
These aren't like the days back when I could self install Linux and expose it to the internet while I configured it.
I’m looking forward to iPadOS. All the heaven from iOS in terms of sandboxing and no background tasks (except very well defined).
Also Catalina afaik will make an os partition which cannot be tampered with.
Which is a very good thing on my book.
Today it is Google's Keystone, tomorrow it is some scummy app downloaded by a grandpa thinking it was a link actually sent by one of his grandsons about their birthday party.