Conjure is an interactive environment for evaluating code within your running program.
The core features of Conjure are language agnostic (although it’s targeted at Lisps for now), with each language client providing their own extra tools. Here’s the currently supported languages, contributions and 3rd party plugins that add clients are highly encouraged! You can find a comparison table for all clients and supported features in the wiki.
-
Clojure over nREPL (quickstart)
-
Fennel inside Neovim via Lua and Aniseed (quickstart)
-
Fennel outside of Neovim within any Lua process through an stdio REPL (quickstart)
-
Janet over spork/netrepl (quickstart)
-
Janet over stdio (quickstart)
-
Racket over stdio (quickstart)
-
Hy over stdio (quickstart)
-
Scheme (MIT by default) over stdio (quickstart)
-
Guile over a socket file (quickstart)
-
Common Lisp over Swank (quickstart)
-
Julia over stdio (quickstart)
-
Rust over stdio with evcxr (quickstart)
-
Lua inside Neovim (quickstart)
-
Python over stdio (quickstart)
You can learn more about the mappings and configuration inside Neovim with :help conjure
. Each language client also has it’s own specific help text under :help conjure-client-{LANGUAGE}-{TRANSPORT}
, such as :help conjure-client-clojure-nrepl
, which contains specific mappings and configuration for that client.
You can also get an interactive guide to the core principals and mappings by executing :ConjureSchool
or by using the "try before you install" script as described further down this page.
What does wielding Conjure actually look like? What can you do with it?
-
YouTube: How I use Conjure and Neovim to interact with Clojure (and more!)
-
Asciinema (see below)
-
Open a file of a supported file type such as
clojure
orracket
. -
Some languages, such as Clojure, will automatically connect to your REPL or give you the commands and mappings to do so yourself.
-
Use various mappings to execute the code you’re interested in such as the current form, file, buffer or even form at a Neovim mark.
-
Every result is kept in a log buffer that you can access at any time.
-
Support for multiple asynchronous autocompletion frameworks, with deoplete built in (
<C-x><C-o>
also works!).-
You can also use nvim-cmp with cmp-conjure. (which I actually recommend!)
-
Conjure allows you to send the right part of your code off for evaluation, see the results, wrap the results in more code and evaluate that. It keeps a log of your actions and results so you don’t have to remember them, that context is extremely helpful when you’re down the rabbit hole of a bug hunt.
Evaluating a form under your cursor is as simple as <prefix>ee
, where <prefix>
defaults to <localleader>
. Be sure to set your your <localleader>
key or configure Conjure to use a different mapping prefix, see :h maplocalleader
and :h g:conjure#mapping#prefix
for more information.
The goal is to give you a tool that lets you effortlessly run whatever you need to while keeping out of your way (but there with context and more information if you need it).
Once installed you can run :ConjureSchool
to begin an interactive tutorial.
You can trial run Conjure with the interactive :ConjureSchool
tutorial without actually installing it! The only prerequisites are curl
and an up to date nvim
.
curl -fL conjure.fun/school | sh
This will temporarily download the plugin, launch Neovim with :ConjureSchool
running and then clean up after itself. You get to try out what Conjure has to offer without having to edit your Neovim configuration.
Requires Neovim 0.7 or newer.
Alternatively you can use Magic Kit, an opinionated starter kit that includes all sorts of essential tools.
use 'Olical/conjure'
Plug 'Olical/conjure'
return { "Olical/conjure" }
To take advantage of lazy loading, a bit more involved configuration is required:
Details
return {
{
"Olical/conjure",
ft = { "clojure", "fennel", "python" }, -- etc
lazy = true,
init = function()
-- Set configuration options here
vim.g["conjure#debug"] = true
end,
-- Optional cmp-conjure integration
dependencies = { "PaterJason/cmp-conjure" },
},
{
"PaterJason/cmp-conjure",
lazy = true,
config = function()
local cmp = require("cmp")
local config = cmp.get_config()
table.insert(config.sources, { name = "conjure" })
return cmp.setup(config)
end,
},
}
All configuration is performed through global Neovim variables, this may change to a .setup(…)
Lua first approach some day but for now you can configure things by setting variables early, before Conjure loads. For example, if you wish to rebind or disable the default documentation lookup key (K
) because you don’t need it or it conflicts with your LSP configuration you can do so like this:
-- Disable the documentation mapping
vim.g["conjure#mapping#doc_word"] = false
-- Rebind it from K to <prefix>gk
vim.g["conjure#mapping#doc_word"] = "gk"
-- Reset it to the default unprefixed K (note the special table wrapped syntax)
vim.g["conjure#mapping#doc_word"] = {"K"}
# And the same disabling but in Vim Script
# Note the v:false in Vim Script!
let g:conjure#mapping#doc_word = v:false
Please see :help conjure
for the full list of possible configuration variables and values.
Modifications or mods are extra plugins that improve Conjure in various ways. They may add completion plugin support, entire language clients or new mappings that do fun and interesting things. You can learn about creating your own by reading the source code of the projects listed below as well as "Using Conjure programatically (API)" and "Client features".
-
jlesquembre/coc-conjure: Conjure completion support for coc.nvim.
-
PaterJason/cmp-conjure: nvim-cmp source for Conjure.
-
Invertisment/conjure-clj-additions-nrepl: Jump to failing Clojure test and more.
-
Invertisment/conjure-clj-additions-vanilla: Jump to failing Clojure test for Conjure when not using nREPL.
Warning! If you want to work with ANY language that isn’t a Lisp dialect you will need to use tree sitter. If you do not use tree sitter only visual selection and vim motion based evaluations will work. You need tree sitter if you wish to evaluate non Lisp languages with
<prefix>ee
and other such form based evaluation mappings.
When you ask Conjure to evaluate the form under your cursor it has to understand the code enough to be able to find the boundaries and extract the right characters from the buffer. This used to be done using Neovim’s built in findpairpos
, syntax highlighting regexes and exhaustive searching of the buffer. This is error prone, gets slow as the buffer grows and doesn’t work with non-Lisp languages which lack clear boundaries.
I highly recommend you set up tree sitter inside your Neovim configuration and :TSInstall [language]
every language you’re interested in working with. You should then keep those tree sitter modules up to date as you upgrade Neovim since the API seems to change slightly over time.
Tree sitter allows you to work with non-Lisp languages like Julia as well as get far more accurate, smart and performant evaluations in languages like Clojure. You can learn more and get everything set up using the nvim-treesitter repository.
It’s technically optional since Conjure contains legacy fallback code, but I highly recommend tree sitter when using Conjure, it’s how you get cool things like smart comment block evaluations in Clojure and form based evaluations in Julia and Lua.
The majority of the documentation can be found within :help conjure
. You can also use :ConjureSchool
to get an interactive introduction to the workflow and mappings Conjure provides. Refer to the list at the top of this file for links to the various quickstart guides for each language.
Broader documentation can be found in the Conjure wiki, there you’ll find blog posts and guides that will help you get common workflows up and running. Contributions are encouraged!
Conjure is written entirely in Lua (no VimL or external processes!) which is compiled from Fennel by Aniseed ahead of time. Check out CONTRIBUTING.adoc
for more on how to work on Conjure using itself.
Historically, Conjure was Clojure specific with an entirely different implementation, you can still find that version on the legacy-jvm branch.
Find the full unlicense in the UNLICENSE
file, but here’s a snippet.
This is free and unencumbered software released into the public domain.
Anyone is free to copy, modify, publish, use, compile, sell, or distribute this software, either in source code form or as a compiled binary, for any purpose, commercial or non-commercial, and by any means.