clojure-lsp
settings are picked up only on server start and can be configured via 4 ways:
clojure-lsp
will look for project specific settings in a file called .lsp/config.edn
. It will search from your project root folder up the directory structure so you can have multiple projects share the same settings.
Example:
{:cljfmt {:indents {#re ".*" ns [[:inner 0] [:inner 1]]}}
:auto-add-ns-to-new-files? false}
For global settings which should work for all the projects using clojure-lsp
, you just need to add the same configs to ~/.lsp/config.edn
or $XDG_CONFIG_HOME/.lsp/config.edn
.
For an example of a global config.edn
, check here.
This is specific for an client, where it sends on startup, check LSP spec for more information. This is useful if you are changing a default for a client/editor that will affect all users of that editor.
This is an example how Emacs lsp-mode
pass custom information.
If you are using a client which defines InitializationOptions as a json object, you can use json types instead:
"incremental"
or ":incremental"
){"unused-public-ns": {"level": "info"}}
)["src" "test"]
)["src" "test"]
)When starting the server, clojure-lsp will search for a specific path clojure-lsp.exports/<group-id>/<artifact-id>/config.edn
for extra configurations in the classpath, only if you set the setting :classpath-config-paths
containing that group/artifact
. Example:
Lib my-org/my-lib
could have a clojure-lsp config edn that makes sense for the usages of that lib, making available on the classpath:
my-lib/resources/clojure-lsp.exports/my-org/my-lib/config.edn
{:cljfmt {:indents {foo [[:block 0]]}}}
and then if your project use/has the my-org/my-lib
inside your classpath, you could add the setting
your-project/.lsp/config.edn
{:classpath-config-paths ["my-org/my-lib"]}
And then clojure-lsp will merge the configuration from the lib with the other configurations from your project.
This is useful if you have some rule to apply to clojure-lsp for multiple projects, mostly using via API for linting for example, and want to move the common configuration to some place instead of adding to each project that needs that.
You can find all settings and its default here and below the docs for each one:
name | description | default |
---|---|---|
source-paths | project-local directories to look for clj/cljc/cljs files, if using deps.edn , project.clj or bb.edn , use :source-aliases instead. | #{"src" "test"} |
source-aliases | Used for deps.edn or project.clj projects, the aliases which clojure-lsp should get the source-paths besides the root level :paths and :extra-paths . Check the source-aliases discovery section below. | #{:dev :test} |
linters | clojure-lsp custom linters, check the diagnostics settings section below | |
additional-snippets | Additional user snippets to be available during completing, check the snippets section below | [] |
ignore-classpath-directories | will not consider clojure files within the directories specified by your classpath. This is needed, for instance, if your build puts artifacts into resources or target that you want lsp to ignore. | false |
lint-project-files-after-startup? | Whether to async lint all project only files after startup to make features like List project errors work. | true |
notify-references-on-file-change | Whether to update diagnostics of the changed references when editing files, avoiding outdated diagnostics in other files. | false |
use-metadata-for-privacy? | Whether to use ^:private metadata for refactorings instead of defn- | false |
:clean | Settings related to clean-ns refactoring. Check clean settings below. | |
semantic-tokens? | Whether to enable LSP semantic tokens server support for syntax highlighting. | true |
hover arity-on-same-line? | Whether to keep the arity on the same line of the function on hover, useful for Emacs users. | false |
hover hide-file-location? | Whether to show the full filename and path on hover. | false |
hover clojuredocs | Whether to get clojuredocs information on hover, the clojuredocs content is cached. | true |
auto-add-ns-to-new-files? | Whether to automatically add the ns form in new blank files. | true |
document-formatting? | if true or not present, document formatting is provided. | true |
document-range-formatting? | if true or not present, document range formatting is provided. | true |
text-document-sync-kind | The sync kind during document changes, if client should send whole buffer or just related changes. Should be :full or :incremental | :full |
dependency-scheme | How the dependencies should be linked, jar will make urls compatible with java's JarURLConnection. You can have the client make an lsp extension request of clojure/dependencyContents with the jar uri and the server will return the jar entry's contents. Similar to java clients | zip |
cljfmt-config-path | Where to find cljfmt configuration for formatting. A path relative to project root or an absolute path. Use #re for regex inside the cljfmt configuration file. | .cljfmt.edn |
cljfmt | If no :cljfmt-config-path is provided, used this for formatting, json encoded configuration for cljfmt | {} |
project-specs | A vector of a map with project-path and classpath-cmd , defining how clojure-lsp should find your project classpath. the project-path should be a file and the classpath-cmd the command to run to get the classpath | Check Classpath scan section below |
code-lens segregate-test-references | Segregate main references from test references with option to disable | true |
classpath-config-paths | List of extra configurations to load from classpath, for more info, check Classpath config paths section. | [] |
cache-path | Where to store the project's analysis cache, used to speed up next clojure-lsp startup. A path relative to project root or an absolute path. | .lsp/.cache |
log-path | A absolute path to a file where clojure-lsp should log. | A JVM tmp path, usually /tmp/clojure-lsp.*.out |
clojure-lsp needs to analyze the whole project and its dependencies to understand your code for most features, during the startup clojure-lsp will try to find the classpath of your project to pass to clj-kondo later.
You can configure how clojure-lsp should find the classpath with the project-specs
setting, check the default here.
Supported project types at the moment are:
leiningen
: If a project.clj
is found at the project root, clojure-lsp will run lein classpath
.deps
: If a deps.edn
is found at the project root, clojure-lsp will run clojure -Spath
.boot
: If a build.boot
is found at the project root, clojure-lsp will run boot show --fake-classpath
.shadow-cljs
: If a shadow-cljs.edn
is found at the project root, clojure-lsp will run npx shadow-cljs classpath
.babashka
: If a bb.edn
is found at the project root, clojure-lsp will run bb print-deps --format classpath
.Note that it's possible to have more and one project type at the same time e.g. deps + babashka, clojure-lsp will merge the classpath and everything should works fine.
Make sure to have these programs available on the PATH
environment variable used by your editor, otherwise clojure-lsp will warn about a classpath scan fail, causing a lot of features to not work properly.
Alternatively, you can configure the project-specs
specific for your project, for example:
.lsp/config.edn
{:project-specs [{:project-path "deps.edn"
:classpath-cmd ["clojure" "-A:dev" "-Spath"]}]}
Note that clojure-lsp will make this scan to save the cache when:
.lsp/.cache
)project.clj
for example) changed.Default: Check :linters
in all-available-settings.edn.
clojure-lsp
uses clj-kondo under the hood to lint the code and retrieve the analysis to
make most of features work, you don't have to install clj-kondo to
make it work.
clojure-lsp
will use a specific clj-kondo version that can be retrieved via clojure-lsp --version
, but make sure you have it
properly configured in your .clj-kondo/config.edn
file.
It has the possible key/values:
:level
with available values: :off
, :on
with default value of :on
:report-duplicates
which will show all linters of the same symbol instead of showing only the first spot. Available values: true
, false
with default value of true
ns-exclude-regex
which will exclude the diagnostics/findings for namespaces that match this regex.Example:
.lsp/config.edn
{:linters {:clj-kondo {:level :on
:report-duplicates true
:ns-exclude-regex "some-ns.*"}}}
If you are a (neo)vim user and have ale installed as a plugin, you
should not have this configured as a linter let g:ale_linters = {'clojure': ['clj-kondo']}
in your vimrc. Having this
linter enabled via ale
will only conflict with the built-in clj-kondo bundled with clojure-lsp.
For more information about all clj-kondo available configurations, check the clj-kondo configuration section
Clojure-lsp register custom linters in clj-kondo, for specifically those linters, configurations should be done on clj-kondo config files, e.g. (<project>/.clj-kondo/config.edn
), below are the custom linters used:
A custom linter that reports public functions/vars not used over the project.
It has the possible key/values:
:level
with available values: :info
, :warning
, :error
or :off
with default value of :info
.:exclude
a whole namespace with #{my-ns}
or a specific var #{my-ns/foo}
.:exclude-when-defined-by
excludes this linter when your var is defined by a macro for example, like #{my-ns/deftest}
.Example:
.clj-kondo/config.edn
{:linters {:clojure-lsp/unused-public-var {:level :warning
:exclude #{my-ns/foo
my-ns/bar
other-ns
my-func}
:exclude-when-defined-by #{my-ns/defflow}}}}
It's not recommended to disable the linter as it provides helpful smart checks/suggestions for your code, even so it's possible via the following config:
.lsp/config.edn
{:linters {:clj-kondo {:level :off}}}
For information on how to troubleshoot the linter, check the troubleshooting section
Some features require know the available source paths of your project, where your code lives, clojure-lsp has some settings for that.
lein
project, clojure-lsp will scan the project.clj
file for :source-paths
, :test-paths
and the optional source-paths
from the specified :source-aliases
setting (default #{:dev :test}
), unless you specified :source-paths
setting manually.deps.edn
, clojure-lsp will scan the deps.edn
file for :paths
, :extra-paths
and the paths
and extra-paths
from the specified :source-aliases
setting (default #{:dev :test}
), unless you specified :source-paths
setting manually.deps.edn
or lein
project, a boot
project for example, clojure-lsp will use only the :source-paths
setting (default #{"src" "test"}
) which should point to the folders containing your clojure code.Default: Check :clean
in all-available-settings.edn.
Where to place first require/import following Clojure Style Guide, :same-line
, :next-line
or :keep
.
Whether to enable sort of ns
children like require,import forms following Clojure Style Guide.
Whether to enable sort of :require
form.
Whether to enable sort of :import
form.
Whether to enable sort of :refer
form.
:max-line-length
: the max refers to keep at same line before breaking the line. Default 80
.Besides the 19 built-in snippets, it's possible to configure custom additional snippets via :additional-snippets
setting:
:name
the name to use while completing to reach that snippet, preferably with a $
sufix to indicate a snippet.:detail
Custom text to show along with the completion name.:snippet
The body of the snippet, besides any text it can contains:
$1
, $2
, ... as the tabstops representing each place where user may change the content.$0
as the last tabstop.$current-form
to replace the current form in the snippet.Example:
{:additional-snippets [{:name "wrap-let-sexpr$"
:detail "Wrap current sexpr in let"
:snippet "(let [$1] $0$current-form)"}]}
when completion is called on the code below with the cursor as |
wrap|(+ 1 2)
It should return a completion item that after applied should result in:
(let [|] (+ 1 2))
Can you improve this documentation? These fine people already did:
Eric Dallo, Noah, Oliver George, Matthew Steedman, David Harrigan, rvlo & dcfrankelEdit on GitHub
cljdoc is a website building & hosting documentation for Clojure/Script libraries
× close