eglot-format-and-organize-imports: format a file and organize its imports

[adonovan]

This function augments eglot-format-buffer so that it in addition to formatting, it organizes the imports too:

(defun eglot-format-and-organize-imports ()
  "Update the imports and format the file, analogous to commands like Go's `goimports`
but agnostic to language, mode, and server."
  (interactive)
  (let* ((actions (eglot-code-actions (point-min) (point-max) "source.organizeImports")))
    (when actions
      (eglot-execute (eglot--current-server-or-lose) (car actions)))
    (eglot-format-buffer)))

You can add it to your favorite mode hook like so:

(add-hook 'go-mode-hook
 (lambda ()
   (add-hook 'before-save-hook #'eglot-format-and-organize-imports nil t)))

In Go, it's the norm for most users to apply these two operations on save, and I assume that must be true in many other languages too. Does this seems like a useful piece of core functionality for the eglot package?

[manuel-uberti]

It may be worth noticing that Eglot already provides eglot-code-action-organize-imports, so a command addressing that and eglot-format-buffer is easier to add to a user's own configuration.

[stapelberg]

I sent https://go.dev/cl/592335 to update our docs regarding the “before-save” terminology.

I don’t understand what you are referring to with regards to eglot-ensure. The Emacs manual clearly documents eglot-ensure without any warnings or discouragement: https://www.gnu.org/software/emacs/manual/html_node/eglot/Starting-Eglot.html

In general, I am disappointed to learn that use-package apparently is contentious among Emacs maintainers. When I first discovered use-package, it has helped me make Emacs start up faster and structure my configuration better. I don’t think telling eglot users to avoid use-package is a good thing :(

[manuel-uberti]

I don’t understand what you are referring to with regards to eglot-ensure. The Emacs manual clearly documents eglot-ensure without any warnings or discouragement: https://www.gnu.org/software/emacs/manual/html_node/eglot/Starting-Eglot.html

There is the following in that manual page, which should be taken into account when debugging problems:

Note that this variant of starting an Eglot session is non-interactive, so it should be used only when you are confident that Eglot can be started reliably for any file which may be visited with the major-mode in question.

In general, I am disappointed to learn that use-package apparently is contentious among Emacs maintainers. When I first discovered use-package, it has helped me make Emacs start up faster and structure my configuration better. I don’t think telling eglot users to avoid use-package is a good thing :(

I think the issue is simply that not everyone uses use-package (I don't, for instance), so taking it for granted when reporting bugs may be a misleading assumption.

[stapelberg]

Note that this variant of starting an Eglot session is non-interactive, so it should be used only when you are confident that Eglot can be started reliably for any file which may be visited with the major-mode in question.

I don’t know what this sentence is trying to convey (it sounds like there might be a problem with error handling?) or what I should be doing instead.

I think the issue is simply that not everyone uses use-package (I don't, for instance), so taking it for granted when reporting bugs may be a misleading assumption.

The way I understood João’s message was that I should not file bug reports if I’m using use-package. I’m not taking anything for granted, nor am I saying everyone should use use-package. I just think using a popular Emacs builtin package should not be a reason to not be allowed to report bugs.

[manuel-uberti]

I don’t know what this sentence is trying to convey (it sounds like there might be a problem with error handling?) or what I should be doing instead.

You should first try with just M-x eglot in the buffer visiting the file you are interested in.

The way I understood João’s message was that I should not file bug reports if I’m using use-package. I’m not taking anything for granted, nor am I saying everyone should use use-package. I just think using a popular Emacs builtin package should not be a reason to not be allowed to report bugs.

Well, if your bug report contains use-package statements and the maintainer(s) (or just a regular user as myself) are not used to this macro, you are asking the maintainer(s) to reproduce a problem with something new to them which they have to learn. It's an extra burden, in my opinion. use-package is popular but that does not mean that everyone should learn it in order to fix bugs.

[joaotavora]

I don’t know what this sentence is trying to convey (it sounds like there might be a problem with error handling?) or what I should be doing instead.

Indeed to make it crystal clear it should say "started reliably with M-x eglot"

you are asking the maintainer(s) to reproduce a problem with something new to them which they have to learn.

Yes, and it's a pretty big language, judging by the docstring and manual. Even if it isn't new, it needlessly obfuscates and makes reproduction harder. You're asking the maintainer's brain to macroexpand those forms. It's like adding details about bathroom decorations when reporting a burglary.