“Anything that can be written in Rust will eventually be written in Rust”
Someone has to explain how rm, which doesn’t allocate any memory (as far as I can tell), isn’t memory safe ?
If I cant remember what dir I’m in, then rm is mot very memory safe is it?
[edit: spelling]
I don’t know whether rm is memory-safe or not, but vpr is. By ‘memory-safe alternative’ I meant that this alternative is memory-safe, but not that rm isn’t.
Reminds me of when they started printing “vegan” and “gluten free” on water bottles.
In GNU coreutils the implementation of rm doesn’t allocate memory however I believe alternative implementations do.
Here’s an example from the OpenBSD source code - https://github.com/openbsd/src/blob/222e275fb89ffb67abe0726dee2b107220092dc3/bin/rm/rm.c#L335
Presumably other *BSDs use something similar? Didn’t check out FreeBSD or anything.
Edit: So I suppose if you are using a BSD-type system (maybe including macOS?), and memory safety was important to you (to the point of extreme paranoia), then you might want to look into this rust project. Or just use the GNU implementation.
Can you share what would be a concrete example of the risk taken by running a RM program with a memory leak or dangling pointers? I fail to see, by my own ignorance, the benefit of memory safety everywhere. But I do enjoy the rust rewrites of shell tools because of the ergonomics, speed, and new functionalities. I’m asking because the first thing you mentioned as a benefit was memory safety.
Unfortunately, I don’t remember the source so we may need to go digging. But I recall reading that something like 1/3 of all bugs are related to memory safety. And those bugs translate to things like buffer overflow and privilege escalation attacks.
The proclaimed advantage is that by making the entirety of Rust memory safe, that entire class of bugs simply won’t exist for projects written in Rust. When they do happen, the bugs will be addressed by the language rather than many thousands of downstream projects. It should be an enormous gain in development performance for the world.
I think the idea makes sense. Time will tell us how well that works.
This probably isn’t the answer you’re looking for, but vpr being memory-safe isn’t a benefit that it has over rm, since rm apparently doesn’t allocate any memory (as @radiant_bloom@lemm.ee wrote).
the first thing you mentioned as a benefit was memory safety.
Looks like I worded my project description poorly. As I wrote in another comment, I meant that this alternative is memory-safe (being written in safe Rust), but not that rm isn’t.
edit: I’ve updated the post’s title to clear things up
The README lacks a description of why I would choose this over rm. The name makes me think it might replace shred but that doesn’t appear to be the case.
To think that something i used to completely nuke my homeserver one time can be written in 112 lines of rust. thats the power of linux right there
Well, I’m not sure how many lines of C rm is written in but I think that rm being only around 4kb (iirc) is something to consider.
But still, storage probably matters least in this day and age. Oh, and…
something I used to completely nuke my home server
If I’m reading this right, then I hope you had backups ready :)
C is better.
