This is the language server implementation for the Elm programming language.
Table of Contents
Note for VSCode users: The plugin contains the language-server. No installation necessary.
The server can be installed via npm (or from source).
npm install -g @elm-tooling/elm-language-server
Then, you should be able to run the language server with the following command:
elm-language-server
You might need to use this, if your using powershell:
elm-language-server.cmd
Follow the instructions below to integrate the language server into your editor.
First, clone this repo and compile it. npm link will add elm-language-server to the PATH.
git clone git@github.com:elm-tooling/elm-language-server.git
cd elm-language-server
npm install
npm run compile
npm link
elm-languager-server and its dependencies are available in nixpkgs.
nix-env -i -A nixpkgs.elmPackages.elm-language-server
You will need to install elm and elm-test to get all diagnostics and elm-format for formatting. Alternatively you can also just install these to your local npm package.json.
npm install -g elm elm-test elm-format
If you want to use elm-review:
npm install -g elm-review
Or use local versions from your node_modules directory, if you want to do that you need to set the paths, via the settings (e.g. set elmPath to ./node_modules/.bin/elm).
We used to have a file called elm-tooling.json where you could specifiy "entrypoints". That’s not needed anymore – the language server finds the entrypoints automatically.
If all you had in elm-tooling.json was "entrypoints", you can safely remove that file.
Currently, no configuration at all is needed.
The Elm Language Server has built-in support for linting. Check out the documentation for configuring the linter.
Supports Elm 0.19 and up
| Feature | Description |
|---|---|
| diagnostics | Provided via elm, elm-test and our own type inference and linter |
| formatting | Provided via elm-format and post-processed to only return a diff of changes. This way it should not be as intrusive as running elm-format normal |
| codeLenses | Currently only shows if a type alias, custom type or function is exposed from that module |
| completions | Show completions for the current file and snippets |
| definitions | Enables you to jump to the definition of a type alias, module, custom type or function |
| documentSymbols | Identifies all symbols in a document. |
| folding | Let's you fold the code on certain Elm constructs |
| hover | Shows type annotations and documentation for a type alias, module, custom type or function |
| linkedEditing | Enables auto renaming a function name when the type annotation name is edited, or vice versa |
| references | Lists all references to a type alias, module, custom type or function |
| rename | Enables you to rename a type alias, module, custom type or function |
| workspaceSymbols | Identifies all symbols in the current workspace |
| selectionRange | Enables navigation by selectionRange (extend selection for e.g.) |
This server contributes the following settings:
elmLS.trace.server: Enable/disable trace logging of client and server communicationelmLS.elmPath: The path to your elm executable. Should be empty by default, in that case it will assume the name and try to first get it from a local npm installation or a global one. If you set it manually it will not try to load from the npm folder.elmLS.elmReviewPath: The path to your elm-review executable. Should be empty by default, in that case it will assume the name and try to first get it from a local npm installation or a global one. If you set it manually it will not try to load from the npm folder.elmLS.elmReviewDiagnostics: Configure linting diagnostics from elm-review. Possible values: off, warning, error.elmLS.elmFormatPath: The path to your elm-format executable. Should be empty by default, in that case it will assume the name and try to first get it from a local npm installation or a global one. If you set it manually it will not try to load from the npm folder.elmLS.elmTestPath: The path to your elm-test executable. Should be empty by default, in that case it will assume the name and try to first get it from a local npm installation or a global one. If you set it manually it will not try to load from the npm folder.elmLS.disableElmLSDiagnostics: Enable/Disable linting diagnostics from the language server.elmLS.skipInstallPackageConfirmation: Skip confirmation for the Install Package code action.elmLS.onlyUpdateDiagnosticsOnSave: Only update compiler diagnostics on save, not on document change.Settings may need a restart to be applied.
| Editor | Diagnostics | Formatting | Code Lenses | Completions | Definitions | Document Symbols | Folding | Hover | Linked Editing | References | Rename | Workspace Symbols |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VSCode | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: |
| VIM CoC | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :x: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: |
| VIM LanguageClient | :heavy_check_mark: | :heavy_check_mark: | :grey_question: | :heavy_check_mark: | :grey_question: | :grey_question: | :grey_question: | :grey_question: | :x: | :grey_question: | :grey_question: | :grey_question: |
| VIM ALE | :heavy_check_mark: | :x: | :x: | :grey_question: | :heavy_check_mark: | :x: | :x: | :heavy_check_mark: | :x: | :heavy_check_mark: | :x: | :heavy_check_mark: |
| Kakoune | :heavy_check_mark: | :heavy_check_mark: | :grey_question: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :grey_question: | :heavy_check_mark: | :x: | :heavy_check_mark: | :heavy_check_mark: | :grey_question: |
| Emacs | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :grey_question: | :heavy_check_mark: | :x: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: |
| Sublime | :heavy_check_mark: | :heavy_check_mark: | :x: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: | :x: | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark: |
Just install the elm-tooling/elm-language-client-vscode plugin from the VSCode MarketPlace
To enable linked editing in VSCode, use the setting "editor.linkedEditing": true.
There are general setup instructions and FAQ for Vim.
It's recommended to install syntax highlighting, which also adds the required detection of elm as filetype. An example vim configuration can be found in elm-vim/vim-config-example.
To enable support with coc.nvim, run :CocConfig and add the language server config below.
If needed, you can set the paths to elm, elm-test, elm-review and elm-format with the elmPath, elmTestPath, elmReviewPath and elmFormatPath variables.
{
"languageserver": {
"elmLS": {
"command": "elm-language-server",
"filetypes": ["elm"],
"rootPatterns": ["elm.json"]
}
},
// If you use neovim you can enable codelenses with this
"codeLens.enable": true
}
Much of this is covered in the Example vim configuration section in Coc's readme.
| Feature | How to use it |
|---|---|
| Diagnostics | :CocList diagnostics |
Configure refresh with "diagnostic.refreshAfterSave": false |
| Formatting | :call CocAction('format') |
| CodeLenses | Requires Neovim. Add "coc.preferences.codeLens.enable": true to your coc-settings.json through :CocConfig |
| Completions | On by default, see Completion with sources for customizations |
|
$ claude mcp add elm-language-server \
-- python -m otcore.mcp_server <graph>