More generally and forgivably, I'd say the IDE must know that two variables of the same name may be different variables, and to know which instances are which vars. Funny enough, this benefit of IDEs might rarely be felt with Javascript due to its uber-dynamic nature (at least without classes).
As soon as you add any kind of metaprogramming or macros to the mix, IDEs start to break, because they can't refactor code that doesn't exist yet. That's an issue for IDEs for C, C++, Rust, D and probably few other languages.
If you're talking about ‘extract method’, any regular imperative language in the C style should have that―possibly as ‘extract function’. I mean also Python, JS, Ruby and the like. The hard part here is to figure out which vars go into the method/function and which of them must be accepted as parameters or returned from the function―that's where the understanding of code structure comes in.
Sure, fiddling with metaprogramming will break IDE's understanding of the code, e.g. in Python, but you can do that carefully and isolate the effects so that the rest of code is still ‘clean’. That's also why Java integrates any and all features into the type and OOP system (afaik), so the code can always be described statically. I guess Haskell is an extreme example of that, but not sure how IDE-friendly it is.
Safe Delete
Extract Method
Extract Constant
Extract Field
Extract Parameter
Introduce Variable
Rename
Inline
Change signature
I'm sure if given enough resources and motivation, it's possible to build superb support for the modern programming languages. For example, you wouldn't claim that Intellij/VSCode/Sublime/Atom has better support for Erlang, for OCaml, for Lisp, for LaTeX, would you?
Emacs evolves slowly. Things take time. And yes, there are always IDEs that do things [slightly] better. At a given moment. Remember, though Emacs outlived many trendy IDEs. Relatively not too long ago IDEs like C++ Builder and Delphi Builder, Eclipse, NetBeans were quite popular. And they did have some remarkable features that Emacs didn't. Do you see any HN posts about those anymore?
Now check, for example, Github Language Stats¹ and find there emacs-lisp. Isn't it surprising to see it somewhere in the middle of the list? Isn't it surprising that it made into that list at all? Remember that it is not a general-purpose language; it is not used in the enterprise; it is not used in the game industry; not suitable for writing mobile apps or web pages. And it also a pretty concise language, it's a Lisp. But there's a lot of it on GitHub alone. Who are these people? Who writes all that code? And how the heck they keep doing it - none of them ever have gotten paid for it, aside from some tiny voluntary contributions and hugely successful Kickstarter campaign for Magit.
If a programming language stays relevant for long enough, people eventually will build emacs packages for it, and they'll keep improving them. But for a proprietary IDE, if it doesn't have a feature you like - you have no other option but to submit a request and wait. Sometimes you have to wait indefinitely.
> People who've been using only Emacs for the last 20 years are in no position to comment on IDEs
I moved to Emacs after almost 20 years of using IDEs. If I am still programming 20 years from now, I'm sure it will be in Emacs.
____
People who've been using only Emacs for the last 20 years are in no position to comment on IDEs