Not that I have understood everything, but even an attempt to follow what MIRI is trying to achieve and how, adds to my fledgling RUST knowledge. Thank you, Ralf!
Great talk. Ralf's blog posts and comments on the Rust forums helped me immensely to understand UB. Now I am much less worried about squeezing every last bit of performance with pointer/bit hacking, as I did when writing C and C++ code, because I realize how dangerous that game is if you misunderstand even a little bit of the rules about UB, and how little of improvement I would get compared to safe Rust, which is already so well optimized by rustc. Talking about Miri in particular, it is a super useful tool and a unique one. And it just so happens that I used Miri for exactly the same purpose mentioned in the end of the talk: to make sure that my library (specifically, a custom bit field implementation) works as intended on a big-endian system. Put it in the Git pre-commit hook and done.
Love the talk and the presenter. Unfortunately the audio is going out of sync (24/25 minutes in, and very slowly) and is completely weird at 26+ when someone in the audience asks a question
![Visualizing memory layout of Rust's data types [See description/first comment]](http://i.ytimg.com/vi/rDoqT-a6UFg/mqdefault.jpg)
Not that I have understood everything, but even an attempt to follow what MIRI is trying to achieve and how, adds to my fledgling RUST knowledge. Thank you, Ralf!
Great talk. Ralf's blog posts and comments on the Rust forums helped me immensely to understand UB. Now I am much less worried about squeezing every last bit of performance with pointer/bit hacking, as I did when writing C and C++ code, because I realize how dangerous that game is if you misunderstand even a little bit of the rules about UB, and how little of improvement I would get compared to safe Rust, which is already so well optimized by rustc.
Talking about Miri in particular, it is a super useful tool and a unique one. And it just so happens that I used Miri for exactly the same purpose mentioned in the end of the talk: to make sure that my library (specifically, a custom bit field implementation) works as intended on a big-endian system. Put it in the Git pre-commit hook and done.
Thanks for sharing this :-)
My goodness, the nitty-picky comments. Can these be suspended until after the talk? Thanks to that gentleman who called it out 28:55
Love the talk and the presenter. Unfortunately the audio is going out of sync (24/25 minutes in, and very slowly) and is completely weird at 26+ when someone in the audience asks a question