I feel like I am witnessing the Rust of Theseus

Agreed. It's a language designed in the 1980s, before any of what R does was fashionable. Some of those design decisions have aged poorly, but not all.

For example, everything is a vector? For statistical programming that is very neat, and it composes well with how data manipulation is done in R.

The way to fix this is to replace the loop body with

closures[[i]] <- (function(x) { force(x); function() x })(i)

So R is lazy, too!

R is great for several reasons. Speaking of spooky action at a distance though, this has bitten me:

closures <- list()

for (i in 1:3) {
closures[[i]] <- function() i
}

print(closures[[1]]())
print(closures[[2]]())
print(closures[[3]]())

These all print 3.

These plots are very interesting. Do you have data to break this down by field or methodology, maybe informed by JEL code?

For C++, Boost covers a lot of terrain as a de facto standard library too, right? Cargo is a more decentralized version of that in a way

If the end goal is to get things done, then a stable language with stable libraries helps, and a broad standard library can go a long way. But the overall ecosystem including documentation and tooling is probably important too. All things that are very expensive to develop

(I count Firefox as a success. Keeping up with two decades of evolving web standards and performance optimization is a feat.)

Then Graydon is a hero for more than one reason.

Mozilla is a little enigmatic. Looking at it from the outside, Firefox and Rust are both extraordinary projects, but it’s unclear if they succeeded because or in spite of the org around them.

Kudos for the apology and for making the effort to grow from it! The contempt culture is real and unfortunately not limited to Rails.

I wonder what happened that made Rust such a clear exception early on. Was Mozilla’s culture and hiring so intentionally humane and that was what shaped it?

I fully agree. But sudo is not in GNU coreutils which is what my originally expressed doubts are about

I have been assuming that coreutils is relatively hard to improve on so performance is not a major motivator. (This is also how I interpreted the post’s remark about priorities.) But I have been wrong before. GNU grep was surprisingly fast until ripgrep got even faster after all.

I’m excited about oxidizing the kernel but I don’t know about coreutils … It does not look like a major attack vector:

security-tracker.debian.org/tracker/sour...

Congratulations, friend! Excited that we might be in the same time zone again :)

Ruby 0.95 was released a little later, in Dec 1995. Based on the changelog, the code base was definitely aware of AIX, HP-UX, IRIX, and Solaris, but also of Linux:

github.com/matiaskorhon...