Re: Renaming eglot -- or at least add an alias?

[João Távora]

Stefan Kangas <stefankangas@gmail.com> writes:

> Hi all, João,
>
> I therefore propose
> renaming it to elsp-mode and using the elsp-* prefix for it.

I'm not sure this is a great idea.

It seems to directly violate two of the items I listed in the plan I
sent some time ago
(https://lists.gnu.org/archive/html/emacs-devel/2022-09/msg01583.html)

* The current GNU ELPA "eglot" package is of type :url.  The plan is to
  simply transition it to :core in GNU ELPA This has obvious advantages,
  such as not needing to change a user configuration to continue
  receiving updates on the same package.

* Another item on that list required that, for a transient period, the
  GitHub hosting can continue to mirror the upstream
  lisp/progmodes/eglot.el file and the GitHub eglot-tests.el can
  continue to work with no changes..

  Also current users a git clone of the GitHub repo will continue to
  work unimpeded.

I think adding one or two aliases wouldn't hurt this directly, but a
"renaming" definitely would.  But then what symbols would you add the
alias to? `M-x eglot`? What about all the other externals ones?
Variables, commands, API symbols?

Even if the fallout from that were somehow acceptable, I think adding
the three-letter sequence "lsp" into the the mix will just fuel even
more confusion between the two Emacs LSP clients. I mean, even with the
current unmistakable "eglot"/"lsp" names, I remember people asking why
some snippet of lsp-mode.el config doesn't work with eglot.el.

I don't think current popular packages such named TRAMP, ido, ElDoc,
Flymake are that much better at describing accurately what they do in
their short names.  Or "Emacs", for that matter.  I think naming has had
little bearing on their adoption.

It's debatable that "LSP" is even a good "descriptive" thing to put into
the name of a package.  For a total newbie to Emacs or to programming,
what even is "LSP", what does it mean?  It's, quite literally, a lower
level implementation detail concerning an implementation protocol.  If
it weren't for the occasional need to shop around for "language servers"
on your system package manager, the user shouldn't even know about it.
If a more complete Emacs distribution, say a "Doom" or a "Bliss" Emacs
would just bundle these server programs, then "LSP" would be totally
meaningless.

I'm not sure any of this would more would-be Emacs maintainers will
start engaging with Eglot maintenance, setting up servers to help
reproduce bugs and understand the architecture of the package.  Those
are the real things needed to propel Eglot adoption, in my sincere
opinion.

João

PS: As you may have guessed, despite being out of this particular loop,
I'm not too hot on the current renaming symbol trend.  I think for
example, having renamed `current-line` to `array-current-line` bring us
little in exchange for the confusion and byte-compilation noise it
introduces. Namespacing problems should be solved with proper
namespacing systems.