Re: Making `eglot-server-programs' a custom variable?

["João Távora" <joaotavora@gmail.com>]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> Why?  This is in a user's init file.  I can more or less see that
>> argument for inter-library dependencies.  But here, w-e-a-load is
>> exactly what's needed.  I use it all the time in my config.
>
> I have 4 uses of it in my ~75kB init file.

Heh, I have it in almost every one of my ~75-750 small elisp files, each
one more or less dedicated to configuring a package that I may or may
not use for 7500 years.

> That doesn't mean we have to find a way to avoid its use, just that it
> would be nice to find such a way.

I think with-eval-after-load sits nicely at the intersection of lazy
autoloads and user Elisp snippets.  Is there another tool that does
this?  use-package, maybe?  But won't that use the same thing
underneath?  Anyway I tend to prefer things that do one thing well and
don't make me learn new DSLs.  with-eval-after-load hasn't let me down.

> "Choose an LSP server" seems to me like a very poor way to customize the
> behavior of the server.
>
> IOW the design of the LSP protocol is not "Emacs-y" enough :-)

Ah.  Yes.  You're probably right about that, how could you not be ;-) ?
I also think it's more about these servers using different backend
strategies themselves.  Last I looked at the C++ server landscape, all
of them provided more or less the same, but consuming different amount
resources.  There's also lots of forking going on.  I think this is
good.

>>> Maybe to reduce the problem we should allow multiple entries per
>>> major mode and use the first that works, without needing to go through
>>> `eglot-alternatives`?
>>
>> Again, why?  Why add more semantics to an already complicated variable
>> when the functional eglot-alternatives plugin works fine?  Furthermore,
>> I'd like to this particular configuration for a future
>> multiple-simultaneous-server-in-one-mode idea.
>
> I'm just suggesting directions.  I don't claim they're the right answer.
> My impression is that this variable is currently both "too complex" and
> "not flexible enough".  Don't get me wrong, it's good enough for now.
>
> But no, I don't have a good replacement to suggest.  My gut just tells
> me that there is a better arrangement.

It's not a very pretty variable, I admit, but it's good enough for now
and could be a lot worse :-)

I once thought it would be a good idea to have a buffer-local variable
eglot-server-program (singular) that major-modes could set as a stronger
suggestion of what to use.  So a future smc++-mode could be based
entirely on tree-sitter for syntax-highlighting and some good C++ server
for everything else, including some LSP-backed smc++-frob-kittens
command that isn't in eglot.el.  The "good C++ server" would be set in
eglot-server-program, called from smc++-mode.  Maybe that "good server"
would even be bundled with the smc++-mode package so that M-x smc++-mode
would probably start Eglot by default.  Can you ask your gut if this is
an acceptable arrangement?

João