I can only approve of the rigorous methodology, only to prove your own preconceptions wrong. On the issues of methodology he calls out himself, my intuition would've been to see if I can grab...
I can only approve of the rigorous methodology, only to prove your own preconceptions wrong.
On the issues of methodology he calls out himself, my intuition would've been to see if I can grab commit messages from a bunch of rust repos and then see if I can use NLP to figure out which of those relate to fixing a build that was broken by a new release of rust. But I'm not in tune enough with rust to know whether that would even be a reasonable expectation. It's also a lot more involved. With that NLP model, I'd then try and see if the relevant commit messages make up more of the bulk over time. Or maybe I misunderstand the point of this exercise as not measuring breakage but rather featurage.
Here is the "I Can't Keep Up With Idiomatic Rust" article, which I couldn't relate to at all. It was also posted here on tildes, but had pretty good discussion on HN.
Here is the "I Can't Keep Up With Idiomatic Rust" article, which I couldn't relate to at all.
It was also posted here on tildes, but had pretty good discussion on HN.
I can only approve of the rigorous methodology, only to prove your own preconceptions wrong.
On the issues of methodology he calls out himself, my intuition would've been to see if I can grab commit messages from a bunch of rust repos and then see if I can use NLP to figure out which of those relate to fixing a build that was broken by a new release of rust. But I'm not in tune enough with rust to know whether that would even be a reasonable expectation. It's also a lot more involved. With that NLP model, I'd then try and see if the relevant commit messages make up more of the bulk over time. Or maybe I misunderstand the point of this exercise as not measuring breakage but rather featurage.
Here is the "I Can't Keep Up With Idiomatic Rust" article, which I couldn't relate to at all.
It was also posted here on tildes, but had pretty good discussion on HN.