Alexis King

Don’t get me wrong, I understand the reasons, and I think this kind of research is super valuable—I thoroughly enjoyed reading this—but at some point you have to wonder about how we’ve utterly failed to fix this mess entirely of our own making. Because this is hard to defend.

I think "documenting natural behavior" is a little more apt than with most languages, since it was 20 years between the first C compilers and standardization. Writing a spec that preserves backwards-compatibility becomes a discovery process on extant compilers

Well, certainly *someone* invented these things, but I merely stumbled upon then and have to discover their properties because no one bothered describing the semantics before go brrrrrrrrr

Yes, I completely agree, as I said here: https://twitter.com/lexi_lambda/status/1281740674933628928 … My point is not that any particular work is bad, just that it can look a bit self-inflicted when you step back and contemplate the curious relationship between programmers, compiler writers, and spec authors wrt C.

I don't agree with the past tense: we are inventing this stuff now.

The problems under discussion clearly come from decisions made in the past we are now having to deal with in the present. That they were *our* decisions and we are free to change them at any time is precisely my point. (Someone should apply the “Who killed Hannibal?” meme to UB.)