Re: Adding major or popular language modes to Emacs distribution

[Eli Zaretskii <eliz@gnu.org>]

> Date: Sun, 29 Aug 2021 04:15:05 +0200
> From: Ergus <spacibba@aol.com>
> CC: Eli Zaretskii <eliz@gnu.org>, hi@ypei.me
> 
> Actually, it is more scalable and optimal to invest time implementing lsp and tree-sitteer building support
> because supporting more external packages and modes one by one for languages that maybe won't be
> popular after some years, so they will become unmaintained and unremovable from vanilla. Most of the code
> in the programming modes (and performance penalty) is actually dedicated to syntax highlight, indentation,
> and code navigation. Something that tree-sitee will do better and faster in most cases. Then language
> specific modes could add just some extra specific details... And we can invest then our time improving, not
> implementing basic support.
> Some external packages are also very limited and there are other external additions and plugins to make
> them functional... So adding more and more modes for languages is worthless. Use a ring to rule them all
> please.

We do have a process of obsoleting unused packages, so modes that
become unused and unmaintained have a clear path there.

Also, please don't assume that LSP and Tree Sitter will be here
forever.  They can also fall out of fashion and die, at which point it
would be good for us to have some fallbacks.

Last, but not least: support for a programming language isn't limited
to what LSP, TS, and their ilk provide, it includes additional
features that are specific to Emacs.  Two trivial ones:
auto-mode-alist and menu-bar items for the language.  Less trivial
ones could include REPL-style invocation, support for language
interpreter, etc.