https://www.youtube.com/live/yPoU4QEv_u8?si=ogS2yrhdfWY8E94E
First off, the speaker Alexander Bluhm, a core OpenBSD developer who coordinates releases and errata at Genua alongside Theo de Raadt dropped some serious truth bombs about security embargos. He took a hard look at why long secrecy windows (like Intel’s months-long silence during Spectre and Meltdown) do more harm than good for open-source ecosystems.
When bugs are kept hidden under strict #NDAs, it completely stalls public code review and breaks automated open-testing pipelines. Worse yet, holding back patches gives users a false sense of security while malicious actors are likely discovering and exploiting the exact same flaws in the dark anyway.
Another fascinating takeaway was his deep dive into why OpenBSD's base system remains strictly built in #C rather than #C++. It all boils down to binary compatibility and seamless patching under the hood. C++ relies heavily on inline functions in header files and aggressive name mangling, which bakes security fixes into binaries at compile time and breaks #ABIs when updated.
By sticking to pure #C, #OpenBSD can ship lightweight dynamic library updates (.so). When a patch drops, every application relying on that library is instantly secured without requiring mass recompilations or breaking user environments.
Note: And before you ask whether Rust would handle this better than C++, #Rust actually brings the exact same problem to the table. Rust doesn't have a stable language ABI, relies heavily on generics and traits, and nearly the entire Rust ecosystem (via #Cargo) favors static linking generating .rlib files and bundling all dependencies directly into the final binary, on top of all the added complexity that comes along with it.
+++
