This doesn’t seem very HTTP/2 specific, though, and it seems more like the conclusion should be “don’t use C for things which matter”. It’s not like HTTP pipelining and keepalives didn’t have plenty of edge cases to worry about.
True, but offering three different protocols with different code paths for the same thing is a lot more attack surface. If you start factoring in different compression methods things look even worse.
The bigger problem is lack of liability: we knew C sucked in the 90s, too, but we had three decades of over-confident dudes saying that was user error and _they_ were smart enough to use C correctly.
Then Rust came along and those dudes either tried to say it was too hard (odd contrast, that) or it’d be too much work to rewrite things.
I’m hoping that LLMs will lower the cost of porting for enough people but what would really help is shedding the idea that the status quo is sustainable. Having consequences for sticking with C would help but we need a way to get companies to pay to migrate without destroying open source.
It's also valid to say that Rust's memory safety is good but the rest of the language sucks. Just as annoying as the over confident c-heads is the rust community that made even countenancing working on an alternative memory safe system a moral sin, which stifled innovation in memory safety. Hell, people barely know that Ada/spark is memory safe (or that sel4 is memory safe C)
Aside from the overhead of borrow checking, what about the rest of the language sucks, in your view?
To me, its type system and traits design make it one of the nicest mainstream languages. I'd rather be using a functional language, like Haskell or an ML, and I do sometimes, but I don't pick my work based on that. I work regularly with Java, Go, Typescript, and Python, and out of those I find Rust significantly superior on several dimensions.
Python is terrible, Go is basically weakly-typed Python, Typescript barely deserves comment other than that it's impressive that such a fancy type system was bolted onto Javascript, and Java is basically COBOL at this point - reliable, but clunky and showing its age. Attempts like Kotlin to leverage the JVM aren't very convincing.
Gleam looks interesting but I haven't had a chance to work with it yet, and it's hardly mainstream.
You mentioned Ada/SPARK, but SPARK is not really for general purpose programming, and Ada in general has a similar issue to Java, but worse because of its age. It's as if someone really liked PL/I but decided it needed more features and a syntax closer to COBOL.
I imagine they are referring to the cognitive overhead of learning to work with it.
Whether that is “something natural to you”, “something you learn and adapt to” or “something you fight with forever” seems to vary per-person, with I suspect, most people falling into the forest 2 categories.
I was talking about coding effort - reasoning about borrowing, lifetime annotations, and the fact that this can make refactoring more involved because lifting arbitrary blocks of code, especially when closures are involved, can require extra work.
Someone who's saying "the rest of the language sucks" may be thinking of something like that, hence my question above.
But when it comes to performance overhead, programming around the constraints of the borrow checker can introduce runtime costs that wouldn't otherwise be necessary - e.g. when cloning is used to handle borrowing conflicts.
Even if, in theory, it's possible to restructure a program to avoid performance impacts, in practice people don't always do that, not least because more efficient solutions may not always be obvious. This comes back to my point about coding effort.
I spent a bunch of time, pre-LLM, to try to replace a bunch of unsafe blocks scattered throughout an ML inference codebase, that someone had used to achieve maximum performance. I wasn't able to get a version that performed as well.
This kind of thing may not matter if you're writing e.g. ordinary web services, but for more performance-critical code, it can be an issue. Of course, `unsafe` is always an option to deal with that, but that undermines Rust's guarantees.
> what about the rest of the language sucks, in your view
I will chime in here to say that the adoption of async was imho a terrible decision and led me to abandon Rust for most purposes. I want proper threads and CSP-based abstraction.
This really isn't true. It's true that those people saying those things existed, but much more recently. Back when all the C/C++ legacy code we depend on today was being written it was the performance penalty imposed by safety that forced developers down that path, not some lack of interest in other options. We were also transitioning from a world where a fair bit of systems code was written in even lower level languages than C (assembler, BLISS, BCPL). The world we're in today effectively began around 1975 when people figured out how to make a C compiler run on an 11/45. All the decisions were pretty much locked in from there on.
> These phrases are not awkward, it's high-quality human English writing.
It’s not high-quality, though — it’s SEO speak to pad blog posts without adding anything of substance. We notice it more because that stuff is so common in the training data but not the experience of anyone who doesn’t spend time on content mills.
The article states the opposite, which I’m almost certain was due to the attacker realizing that Content-Security-Policy was blocking the payload on many sites:
“A full inline copy of the payload, which runs through eval if 1688360.cc is blocked.”
> The article states the opposite, which I’m almost certain was due to the attacker realizing that Content-Security-Policy was blocking the payload on many sites:
>
> “A full inline copy of the payload, which runs through eval if 1688360.cc is blocked.”
No, the article doesn't state that the JS served by all listed first party websites was compromised (but see note below). The original maliciously modified content was still (in most cases) served from an optimizely domain like cdn.optimizely.com. The cdn.optimizely.com would not have been blocked by web browsers. The "full inline copy of the payload" refers to content served through that optimizely payload, instead of those additional domains the attacker had set up. About a paragraph or so above your quote:
> The file was still served from cdn.optimizely.com, with Optimizely's certificate, from a host that every affected site already trusted.
That said, while this is a very common scenario, it doesn't cover all of them. There's still a possibility that certain optimizely customers have set up a reverse proxy for the optimizely JS snippet. If this proxy was through a separate sub-domain it's possible NoScript would have caught it (if user opted for sub-domain level blocking). If this proxy was through one of their own primary domain names that user already trusted in Noscript, it would not catch it. So it's not perfect, but covers most common scenarios.
Taking a step back, I'm not sure why there is so much pushback in this thread. Most of it seems to be based on either no or incorrect reading of the information or incorrect understanding. I'm not in any way associated with NoScript or trying to convince folks to use it. Given the incident, I think it's worth stating that this type of attack falls under its main use cases. A reasonably configured NoScript would have protected users in most of these cases. Whether it makes sense for users to use it or not is up to them - it depends on where you see the line between security and convenience tradeoff.
> A reasonably configured NoScript would have protected
I don't doubt that. But how many users can achieve a reasonably configured NoScript, without breaking things they want to use? I am a developer, so I should be able to do it or I were a bad one. But I am sick of an immoral industry that requires me to do extra work because they are trying to steal something from me in an ever growing multitude of ways every time I click somewhere and even without.
It could happen but if you look at the last few decades of losses from software bugs, it’s exceedingly rare for the cost to be more than lost future business, and even that’s unusual. Remember when CrowdStrike brought down businesses around the world and they suffered … having their share price more than double?
They didn’t used to but now they’ll actively attract suspicion if they try to stay out of social media entirely. Business people taking pictures at meals, when touring partners’ events, etc. is the norm now in a lot of fields to the point that it’d be odd if you were doing an executive search and didn’t find at least some vacuous posts on LinkedIn.
This is one of the reason why the US OPM hack was so bad for American intelligence: spies need to blend in on social media but once a face is linked, it’s burnt for anything in person.
I’m also reminded of the clowns pushing “it’s smaller!” container images based on social media vibes rather than analysis where, yes, not using Debian+glibc saved tens of megabytes on the base OS but did so at the expense of measurable performance reductions & compatibility issues like the one you mentioned, and literally every single one of those dudes I worked with was shipping containers pushing a gigabyte so the size reduction had no measurable impact on deployments.
Remember that the non-defense sector of the government has been having cuts for decades and lots of important work is outsourced to rotating contract teams. This is definitely unacceptable but I'd bet there were more than a few people there who were profoundly unsurprised that the backlog caught up with them.
You might be detected but if, for example, you were smart enough not to MITM the same client twice after an error, I suspect a ton of people would ignore it and just assume the airport/coffee shop WiFi was flaky.
I’d also bet more things use libcurl behind the scenes than we appreciate, and that at least a few of them have things like silent updates which retry after a failure without notifying a user. I’ve seen a lot of agents which just try to periodically check in because they’re designed not to be noisy on lousy networks.
reply