I agree that it's a tough situation. "The type system guarantees memory safety" is an extremely important pillar of Rust's identity. They kind of have to portray all soundness issues as "more compiler bugs than something broken in the language itself" (see eg https://news.ycombinator.com/item?id=21930599 which references a GitHub label that AIUI would've included the trait coherence thing at the time) to keep making that claim. It is a core part of the marketing strategy.
Yes, so there's a few things going on here: the first is, I absolutely pattern matched on the cve-rs link. Most people bringing that up are trying to bring up a quick gotcha. I did not follow the first link, I assumed it was to that. I am not educated on that specific bug at all.
I still ultimately think that the framing of Rust being any different than other languages here is actively trying to read the worst into things; Rust is working on having a spec, and formally proving things out. This takes a long time. But it's still ongoing. That doesn't mean Rust marketing relies on lying, I don't think most people even understand "soundness" at all, let alone assume that when Rust says "there's no UB in safe code" or similar that there's a promise of zero soundness bugs or open questions. That backwards incompatible changes are made in spite of breaking code at times to fix soundness issues is an acknowledgement of how sometimes there are in fact bugs, this doesn't change that for virtually all Rust users most of the time, updating the compiler is without fanfare, and so in practice, it is backwards compatible. I have heard of people struggling to update their C or C++ compilers to new standards, that doesn't mean that those languages are horribly backwards incompatible, just that there is a spectrum here, and being on one side of it as close as realistically possible doesn't mean that it's a lie.
But, regardless of all of that, it does appear that the issue you linked specifically may be not just a bug, but a real issue. That's my bad, and I'll try to remember that specific bug in the future.
> They kind of have to portray all soundness issues as "more compiler bugs than something broken in the language itself" [] to keep making that claim.
That's part of the "interesting question" I referred to in my comment. There's probably multiple factors that go into the decision of what to put onto the front page, and the presence/absence of soundness issues is just one of those factors.
