[kbenson]

It's unfortunate that our profession emphasizes ease of learning so much. There are things you can make better to make a language easier to learn that don't affect usefulness, but at some point those mostly run out and there is a trade-off between making something easier to learn and easier to use. One brings more new people into the community, the other retains them by continuing to provide good value for invested time.

Python is actually a good case for this, where choices to make the language simpler and more uniform led to knock-on effects that ultimately lead to aspects of it being quite cumbersome (IMO, obviously). For example, how anonymous functions are specified, how cumbersome and limited that makes map and grep, and how that leads to list comprehensions, which for all but the simplest cases are much more complicated to read and write than a simple chain of maps and greps (assuming something better than lambda).

At the other end of the spectrum you have APL and its relatives. Lots of work to learn and become proficient in, but then you can read and write terse, performant and versatile programs with it quickly. Of course, the pool of people that can understand what you wrote is orders of magnitude smaller than with some of the easier to learn languages.

I understand the urge to make Rust as accessible as possible, and I applaud that effort. I just hope that Rust continues to walk the line of accessibility without compromise on the core ideals.

[Manishearth]

I don't think the intent is to compromise on these :)

But lots of things can be made easier to learn via better tooling, better docs, or better diagnostics. Or just removing papercuts.

One such thing I've recently been thinking about is https://github.com/Manishearth/rust-clippy/issues/1737 , which would be cool if it got big.

[kbenson]

Oh, that would be nice. :)

Rust, being largely of new things (in implementation and and presentation, if not theory), has plenty of room currently to explore changes and features that have little or no negative aspects. I'm a little worried about what happens when the low hanging fruit is fine though.

That said, rust has been very good so far at assessing the merit of proposed changes. I recall a few months back that reasoning was explained, and it assessed positive and negative aspects on a few well defined axes.