So it looks like the #sbcl project is getting sloppified. Very, very saddening to see. It feels like something is slipping through my fingers, so I thought I should say something publicly.

Evidently Christophe is not interested in shutting this crap down, and when I read his email about it, I pretty much gave up on saying something myself. His stance is hopelessly naïve, and this is only being shown now.

As soon as you get these idiots a gap, they'll wedge it open, and it's already happened.

@shinmera but I have not considered the qUaLiTy of the incredibly heisenbug-prone code
@shinmera seriously I don't trust anything short of overthinking it yourself, because I myself managed to land a bug which took two weeks to trigger using my own programs as stress tests
@shinmera there's at least two inaccuracies in this message https://sourceforge.net/p/sbcl/mailman/message/59301540/ (it's definitely still UB to peek at one field given a reference to another, and you can definitely build SBCL on -O0) but I don't feel like maintaining my own SBCL fork
[Sbcl-devel] [PATCH] use memcmp in page_extensible_p for well-defined type+gen comparison | Steel Bank Common Lisp