3 ms·
I think that TL;DR misses lots of nuance; I'd suggest a better TL;DR would be "the features and idioms of different languages can lead to different preferred so
by kibwen 8y ago
I think that TL;DR misses lots of nuance; I'd suggest a better TL;DR would be "the features and idioms of different languages can lead to different preferred solutions to the same problems, which means that we should be careful not to unnecessarily restrict our perception of new languages in terms of the familiar idioms of old ones." Or maybe: "performance benchmarks are hard, and even carefully-crafted comparisons can miss the point if they paper over important practical implementation concerns; an apples-to-apples comparison can unintentionally mislead if it precludes the consideration that perhaps what you actually wanted was an orange after all". Importantly, the article explains why Rust is faster on this specific benchmark, but goes out of its way to note that benchmarks should always be interpreted with appropriate caution, and that nobody should come away with the notion that this implies any sort of strict dominance of Rust over C.
- igouy 8y agoSo it's called "The relative performance of C and Rust" to put the emphasis on the data structures? :)
- kibwen 8y agoAs suggested in the introduction, one would think it was called "the relative performance of C and Rust" because this article is a corollary to a previous article and conceived as a response to an inquiry regarding that previous article--specifically, to answer the question "in your given benchmark, what explains the relative performance of C and Rust?" :P The basic answer: "data structures"; the insightful answer: "data structures that were originally chosen to play to the different strengths and weaknesses of their respective implementation languages, giving us food for thought as to how language design can shape the solutions that users of that language will tend to arrive at, and the performance implications thereof".
- igouy 8y agoObviously Why Rust BTreeMaps beat C AVL Trees is way too-long and misleading a title ;-)
- kbenson 8y agoJust because that would indeed focus on the specific sub-portion of the article that you think is most important, doesn't necessarily make it better than a title that meant to encompass the issues surrounding whole act of benchmarking, specifically between languages that might encourage different idiomatic solutions, and then drills down to what that means in this specific instance. In other words, what you took away from the article might not have been what other people did, or what the author meant to express. It surely wasn't what I thought was important (but neither was "rust is faster/slower than C").