I am looking for a webpage that I could give to colleague to give them a rough idea of what Rust is and isn't, what you should expect (tooling, learning curve, libraries, …), why you should and should not use it,… It should be both complete, and digest, so probably something that take 10mn to read/watch to have enough substance, while not being too intimidating.
The homepage of https://www.rust-lang.org/ is way to succinct, and looks more like a commercial than an honest presentation. I was surprised that it wasn't linked to something more detailed.
I think that starting using Rust for our project at work is a really good idea, but if I speak myself, I look like an evangelist witch isn't a good way to present it. Our team consists mostly of scientist and not programmers, so the readability, quality of the errors, and maintainability by non-programmer is really important. I think that Rust cross all the marks, but I would have like to have a more neutral presentation to handle them.
I think it's a delicate balance. Start pushing Rust too hard, or anything really, and people think you are a fanatic and will push back more.
One of the best introductions to the pros and cons or Rust I have ever seen, without getting into too much technical detail, is this presentation by Jon Gjengset: "Considering Rust" Considering Rust - YouTube
It's beautifully calm, rational and balanced. I think Jon has the credentials to talk to scientists.
I worry that they are not programmers though. I imagine Rust is a bit of a challenge for a rank beginner.
Scientist will not modify the code a lot, but when they do, they should not destroy everything (and god, they are really competent at that!). I personally like the idea that rustc will warn you each time they try to do it, but they may not like it!
Correctness and repeatability.
Not that much strangely. The only one that high value those is the one that know how to program.
Well, there you go. Rust is very resilient to being messed up when refactored and modified. Yes, it will drive them crazy. It's good for them.
I don't move in scientific circles but I have read a lot of stories in the past year or so about results in scientific papers being unrepeatable due to lost source code and so on. I thought this had become a general concern.
I'm dealing with this right now. I'm trying to reimplement a paper's algorithm, but they don't say how they pruned the search tree in their branch and bound, and I'm nowhere near being able to search these binary trees with 2⁷⁵ nodes, and I'm capping it at a depth of 25, which unsurprisingly gives poorer results.
