Conversation
Notices
-
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 07:14:26 JST Sick Sun youtube commenter is more worried about supply chain attacks than memory safety - Haelwenn /элвэн/ :triskell: and Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧 like this.
-
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 07:16:18 JST Sick Sun what if I told you that the existence of supply chain attacks doesn't absolve you from competent programming -
Embed this notice
mangeurdenuage :gnu: :trisquel: :gondola_head: 🌿 :abeshinzo: :ignucius: (mangeurdenuage@shitposter.world)'s status on Wednesday, 04-Dec-2024 07:18:15 JST mangeurdenuage :gnu: :trisquel: :gondola_head: 🌿 :abeshinzo: :ignucius: @sun Both are as worrying. I'm worried of common points of failure like hardware/firmware bugs/backdoors. -
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 07:26:04 JST Haelwenn /элвэн/ :triskell: @sun Well except memory safety is a language feature, not a programmer skill, although you do have to pick the right language for what you need.
While dependency issues how is that not a skill issue? -
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 07:34:56 JST Sick Sun @lanodan right but just imagine being like I don't think I need memory safety because the real risk is supply chain -
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 07:44:55 JST Haelwenn /элвэн/ :triskell: @sun Which is a dead argument because good luck quantifying software risks in such a broad manner, there's barely even have proper infra for communicating security issues to downstreams (like CVEs do) so stats would be seriously off.
Although I think you could say that not all software benefits from memory safety (like when there's little to no external input) and that all software benefits from taking dependency issues seriously. -
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 07:47:03 JST Sick Sun @lanodan that is the difference between "all" and "almost all" -
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 07:48:51 JST Haelwenn /элвэн/ :triskell: @sun And presenting logic to apply instead of "trust me bro, you do (not) need this" in a very general manner. -
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 07:49:31 JST Sick Sun @lanodan lack of memory safety is actually bad but I admit it's hard to avoid C -
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 07:56:31 JST Haelwenn /элвэн/ :triskell: @sun Yeah in fact kind of fun that rust stdlib just links to the libc, rustc depends on LLVM (so C++) and cargo has quite a lot of dependencies in C and various kinds of assembly.
At least Go is more isolated there but for many years reference Go toolchain was in C. -
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 08:01:19 JST Sick Sun @lanodan Rust unsafe still benefits from type safety and borrow checker is not disabled -
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 08:12:43 JST Haelwenn /элвэн/ :triskell: @sun Sure but Rust Unsafe is basically Rust with few specific safety guarantees disabled.
Meanwhile issue with linking to C++/C/assembly/… is you do not have memory isolation, that's between processes.
For example Rust code linking to OpenSSL-Heartbleed is just as exploitable as C code would. -
Embed this notice
Sick Sun (sun@shitposter.world)'s status on Wednesday, 04-Dec-2024 08:13:56 JST Sick Sun @lanodan it doesn't make you perfectly safe it makes you more safe
it also doesn't solve logical vulnerabilities -
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 08:14:54 JST Haelwenn /элвэн/ :triskell: @sun > it also doesn't solve logical vulnerabilities
I know, see the vulns Pleroma had, Erlang and Elixir are memory-safe. -
Embed this notice
frogzone@wizard.casa's status on Wednesday, 04-Dec-2024 08:34:03 JST frogzone @mangeurdenuage @sun yeah they are both bad, just look at what isreal did to lebanon, people dont expect their device to act as a chemical weapon but that line is now crossed.
-
Embed this notice
Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 04-Dec-2024 08:34:04 JST Haelwenn /элвэн/ :triskell: @sun Or for another example on glibc systems, CVE-2023-4911 (glibc ld.so GLIBC_TUNABLES) would still work against sudo-rs instead of sudo.
Meanwhile a Go reimplementation of sudo could end up fine if cgo isn't used.