> It’s also heartbreaking to think about how many jobs are so straightforwardly good for the maker and good for the buyer but economically unviable because of the world we live in. I feel like the world would be a lot different if the cost of living was lower.

💯

Honestly all survey papers should include a graphical taxonomy of related work.

Next you’ll be telling us that lexical scope isn’t that expensive!

It's even responsive!

Interesting, thanks! Apparently it comes from Modula-3.

(Just noticed that the Kotlin doc shows assigning into a temp var, yeah I don't use it that way.)

Sure, I use assertions as well (depends on the situation). But I find checkNotNull very handy for stuff like this:

const newCount = checkNotNull(refCounts.get(node)) - 1;

I generally wouldn't assign the result into a new var, I agree that an assertion or type predicate is preferable then.

Yeah, I'm not sure either. I think the argument is that due to the conversion rules in C++, you aren't (effectively) making the invalid states unrepresentable.

So I buy the argument for C++, but it's less clear to me that it makes sense for languages like Mojo or Odin.

Bjarne Stroustrup: "Subscripts and sizes should be signed" open-std.org/jtc1/sc22/wg...

Thanks for writing/sharing this! Definitely interesting, and you're writing is always a pleasure to read.

This seems like such a tiny thing but I always thought using unsigned ints for sizes was the "correct" thing to do, so I do it even though it's generally a little annoying (b/c casts, fretting about whether temp vars should be unsigned or signed, etc.)