Every language, including all the languages described in this article, goes through a period of instability while it figures out what works and what doesn't.
Author's point was that we should not use "not good" languages for the fear that we might be stuck with them for next 20 years. I'd rather be stuck with a language whose designers are very resistant to change vs one that gets features haphazardly bolted on every few years (PHP comes to mind).
I absolutely agree! However, I don't think Rust will continue to go through wild changes for much longer. My guess is that it will settle and become pretty fixed.
And Haskell certainly doesn't introduce breaking changes very often.
We're shooting for the end of the year, in fact. You can view the list of backwards incompatible language changes here: https://github.com/rust-lang/rust/issues?labels=P-backcompat...
Actually, I'd say instability is one of the significant challenges with adopting Haskell for long-lived production code. For example, there have been a few discussions in various forums and blogs recently about how much of Real World Haskell no longer even compiles on the latest GHC and current versions of libraries. RWH is a book that rapidly became the go-to text for new Haskell developers only a few years ago, so we're not talking about either bleeding edge functionality or a length of time where software written back then has probably been retired here.
So stability is perhaps an area where Go does have an advantage over the likes of Rust and Haskell today. Rust is still evolving as a new language inevitably will; it has not yet reached the level of stability needed for long-term production use by the general programming community. Haskell is also still evolving, but for a different reason: it is valued as much for being a test bed of bleeding edge programming language design as it is for being a practical programming language.
Perhaps Haskell the core language doesn't but ghc certainly does, in every major version.
Every single thing listed in this article could be added to TI BASIC at any point. It would just require a complete re-formulation of the language into something completely unrecognizable.
The <> notation in id<T>(item for generic bracketing is harder to read than other bracketing symbols, e.g. () [] and {}. Unlike those others, angles are used for comparison ops and arrow heads also. If Go ever introduces generics, the Scala-like [] notation looks cleaner and would fit into Go's existing grammar.
type Foo[T] T
type Foo []int
type Array[T] []T