| |
|
monkyyy 
Posted in reply to Guillaume Piolat
| On Friday, 22 August 2025 at 20:47:19 UTC, Guillaume Piolat wrote:
> D evaluation order at CT isn't documented, and is pretty much inconsistent.
I find it predictable, not intuitive, but literally predictable. I dont think theres chance its changeable given the number of order dependent bugs that exist and O(N^N^N) chances it only happens to work in existing code by trail and error where N is lines of code in the industry code bases which I hear is high
I think this is why d compiles fast compared to c++ and rust
I hope all'y'all upstream change your policy on this and just teach it somehow, spec-ify it and give basic tools for brute forcing it
|