having two very different code paths for measuring the length of the DER buffer and writing the DER sounds very scary. i guess its fine with Rust but the idea would give me the heebee-jeebies for any other language unless they are using a safe buffer implementation. i would find it hard to trust that there is no buffer overflow based on divergent behaviour between the two functions.
The post title is misleading and the content reads more like a guerilla advertisement for claude. TL;DR: author works for Anthropic, and used claude to implement an optimization for LLVM.
having two very different code paths for measuring the length of the DER buffer and writing the DER sounds very scary. i guess its fine with Rust but the idea would give me the heebee-jeebies for any other language unless they are using a safe buffer implementation. i would find it hard to trust that there is no buffer overflow based on divergent behaviour between the two functions.
The post title is misleading and the content reads more like a guerilla advertisement for claude. TL;DR: author works for Anthropic, and used claude to implement an optimization for LLVM.