LmCast :: Stay tuned in

Lisp in Vim (2019)

Recorded: May 23, 2026, 3:58 p.m.

Original Summarized

Lisp in Vim - Susam Pal

Lisp in Vim
By Susam Pal on 07 Dec 2019
Introduction

Fifteen years ago, writing Lisp code in Vim was an odd adventure.
There were no good plugins for Vim that assisted in structured
editing of Lisp s-expressions or allowed interactive programming by
embedding a Lisp Read-Eval-Print-Loop (REPL) or a debugger within
the editor. The situation has improved a lot since then. In the
last ten years, we have seen active development of two Vim plugins
named Slimv and
Vlime. Slimv is over
10 years old now. Vlime is more recent and less than 3 years old
right now. Both support interactive programming in Lisp.

I am going to discuss and compare both Slimv and Vlime in this
article. I will show how to get started with both plugins and
introduce some of their basic features. I will not cover everything
though. This is not a tutorial. For tutorials, see
the References section.

If you are looking only for a comparison of the two plugins or a
quick recommendation, jump directly to the
Comparison of Slimv and Vlime
section or the
Quick Recommendation
section.

Contents

Introduction
Background

Lisp
Emacs: SLIME
Vim: Slimv/Vlime

Vim Plugin Management
Software Versions
Get Started

Get Started with Slimv and SBCL
Get Started with Vlime and SBCL
Get Started with Paredit

Use Debugger and Inspector

Use Debugger and Inspector with Slimv
Use Debugger and Inspector with Vlime

Trace Function

Trace Function in Slimv
Trace Function in Vlime

Nifty Features

Evaluate Top-Level Form
Rainbow Parentheses
Argument List
Omni-Completion
Describe Symbol
Expand Macro
Cross Reference

Other Common Lisp Implementations

Use Slimv with CLISP
Use Slimv with ECL
Use Vlime with CLISP
Use Vlime with ECL

Other Lisp Dialects

Use Slimv with MIT/GNU Scheme
Use Slimv with Clojure

Comparison of Slimv and Vlime
Quick Recommendation
Disclosure
References

Background

Before we get started with Slimv and Vlime, it would be nice to take
a brief look at the heritage behind these plugins. These plugins
provide Lisp development environments for Vim, so their story begins
with Lisp.

Lisp

Lisp is a family of programming languages with a distinctive, fully
parenthesised prefix notation. It is quite unlike most of the other
popular programming languages today like C, Python, Ruby, etc. Its
homoiconic nature and its powerful macro system that can transform
arbitrary Lisp expressions make it such a flexible, versatile,
extensible and introspective language that articles describing Lisp
often have the word "enlightenment" in them. For example, see the
following articles:

Beating the
Averages (by Paul Graham)

The
Nature of Lisp (by Slava Akhmechet)

How
Lisp Became God's Own Programming Language (by Sinclair
Target)

Lisp has been described in various ways by various eminent
personalities in the history of computing. Alan Kay has famously
described Lisp as:

The greatest single programming language ever designed.

John Foderaro has written this about Lisp:

Lisp is a programmable programming language.

Eric S. Raymond has expressed the enlightenment one experiences by
learning Lisp in his famous article titled
How To Become A Hacker:

Lisp is worth learning for the profound enlightenment experience you
will have when you finally get it. That experience will make you a
better programmer for the rest of your days, even if you never
actually use Lisp itself a lot.

Randall Munroe, the creator of the
XKCD webcomic has dedicated two
comic strips to Lisp:

Lisp by Randall Munroe
(Source: https://xkcd.com/224/)

Lisp Cycles by Randall Munroe
(Source: https://xkcd.com/297/)

Developed in 1958 by John McCarthy, Lisp is the second oldest
programming language in use today. Only Fortran is older, by one
year. Some of the popular Lisp dialects today include Common Lisp,
Scheme and Clojure. Most of this article would focus on Common
Lisp. Scheme and Clojure would be discussed briefly towards the end
of this article.

Emacs: SLIME

Many Lisp programmers immediately think of Emacs when they think of
writing Lisp code. Emacs is a family of text editors. An Emacs
editor itself is typically implemented in a dialect of Lisp. There
is an Emacs mode named SLIME that provides excellent support for
programming in Lisp. SLIME stands for Superior Lisp Interaction
Mode for Emacs. First released in August 2003, SLIME was created by
Eric Marsden and then later developed further by Luke Gorrie and
Helmut Eller. It offers a Read-Eval-Print-Loop (REPL), integrated
debugging and interactive evaluation of expressions, all available
right within the editor. There are several nifty key bindings
available to compile and evaluate parts or whole of the code in the
current buffer.

SLIME works by launching a Swank TCP server. Swank is a backend
server program written in Common Lisp that listens on a socket to
receive SLIME commands from Emacs and execute them. SLIME is so
useful that it is considered to be indispensible by many Lisp
programmers who write Lisp code in Emacs.

Vim: Slimv/Vlime

Is there anything similar to SLIME for Vim? Yes, there are two
popular options:

Slimv: It stands for Superior Lisp Interaction
Mode for Vim. It is a Vim plugin created by Tamas Kovacs that
was first released in January 2009.

Vlime: It is a Vim plugin created by Kay Z that
was first released in May 2017. It is much more recent than
Slimv. Vlime is younger than Slimv by eight years.

Both plugins use a client-server architecture like SLIME does in
Emacs. Both plugins rely on Swank server to be started. In fact,
Slimv bundles a slightly modified version of Swank with it, so that
it can launch it and connect to it to send expressions to be
evaluated. Vlime does not bundle Swank server with itself but it
provides a wrapper that automatically downloads Swank server when
needed.

Vim Plugin Management

When I started using Vim fifteen years ago, we used to just download
a Vim plugin and copy/extract it to ~/.vim directory.
These days, there are a few plugin management tools for Vim such
as Pathogen,
Vundle,
vim-plug, etc. I
am not going to use any of them because I don't know which one of
them you use and I don't want to write down steps for each one of
them.

In fact, I have never used any Vim plugin manager myself. Until Vim
7, I used to create a ~/.vim/bundle directory, then
copy each plugin to its own directory within it and add the plugin's
directory path to Vim's runtimepath option.

Vim 8 has native support for packages which makes installing plugins
and loading them simpler. With Vim 8, we can copy each plugin to
its own directory within ~/.vim/pack/plugins/start and
they are loaded automatically when Vim starts. The directory name
plugins in this path is only an example. It could be
any arbitrary name and Vim would still load the plugins fine.

In this article, I will use Vim 8's native support for packages to
set up Vim plugins. The only exception to this would be installing
Vlime. The top-level directory of Vlime is not the plugin
directory. The plugin directory is contained in a subdirectory
named vim. This does not conform to the directory
structure of plugins in a Vim package. Therefore, in this article,
I will set up Vlime in the old fashioned way by copying it
to ~/.vim/bundle and then adding the path to its plugin
directory to Vim's runtimepath option.

Software Versions

Since this article involves several layers of software, some of what
is written here may not hold good in future as these various pieces
of software change and evolve over time. Therefore, in this
section, I will note down the versions of various software tools I
used while writing this article. Here they are:

Debian GNU/Linux 10.1 (buster)
Vim 8.1

Slimv (Git repo last updated on 30 Nov 2019 with commit
47a0070)

Vlime (Git repo last updated on 16 Oct 2017 with commit
065b95f)

Paredit (Git repo last updated on 30 Nov 2019 with commit
d99905a)

Rainbow Parentheses (Git repo last updated on 29 Oct 2015 with commit
27e7cd7)

SBCL 1.4.16.debian
GNU CLISP 2.49.92
ECL 16.1.3
MIT/GNU Scheme 9.1.1 on Debian GNU/Linux 9.11 (stretch)
Clojure 1.10.1
Quicklisp beta (libraries last updated on 30 Nov 2019)
tmux 2.8

You will probably need only a tiny subset of the tools above
depending on which sections in this article you would follow. Just
pick the sections you want to try out and follow the steps written
in them. They will walk you through the procedure to install the
tools applicable to the sections you have picked. Regardless of
which sections you pick, I recommend that you definitely go through
the three "Get Started" subsections below. These subsections go
into detail about some of the prerequisites such as support for
tmux, Paredit, support for Python interface in Vim, etc. that are
not covered in the other sections.

The choice of Debian may look like an odd one. I want the commands
and steps discussed in this article to be easily reproducible in a
free and open source operating system. Debian happens to be my
favourite. What works on Debian is easily reproducible on Ubuntu
and other derivatives, often without any changes to the steps. I
believe, it will not be too difficult to translate the steps
provided for Debian to the steps that would work on another
operating system.

Note that Quicklisp (a prerequisite for Vlime) is still beta
software at the time of writing this article. The actual steps to
install Quicklisp may change in future. Check
https://www.quicklisp.org/
for the most up-to-date instructions to install Quicklisp.

Get Started
Get Started with Slimv and SBCL

Here are the steps to set up Slimv and use it:

Install the tools required to set up a Common Lisp development
environment with Slimv with this command:

sudo apt-get install vim-nox sbcl tmux git

The default Vim in Debian is vim.basic provided by
the vim package which does not have support for
Python interface. Slimv is written in Vim script, Lisp and
Python 3, so it does need a Vim package that has support for
Python interface. One such package is vim-nox that
provides the vim.nox command. Installing it
automatically updates the vim command to
run vim.nox. Another such package
is vim-gtk which additionally provides GUI support.
The graphical Vim known as GVim can be launched with
the gvim command. It runs in the desktop
environment. For the purpose of this article, I will stick
to vim-nox because it is lightweight. All steps
meant for Slimv would run equally well on vim-gtk,
MacVim and GVim.

Installing tmux is optional. Slimv can launch Swank server
automatically if Vim is running within tmux, GNU Screen or a
desktop environment, so if you are using GNU Screen already, you
don't need to install tmux. Also, if you are running Slimv in a
desktop environment, you don't really need to install either
tmux or GNU Screen, although you could if you would like to see
Swank running in a separate tmux or GNU Screen window rather
than a separate terminal window. In this article, I am going to
assume that Vim is running within tmux.

If you are going to run Slimv in a terminal without a desktop
environment, tmux or GNU Screen, Swank server has to be run
manually. Point 4 below explains how to do it.

Installing Slimv is pretty simple. Here is one way to do it:

git clone https://github.com/kovisoft/slimv.git ~/.vim/pack/plugins/start/slimv
vim +'helptags ~/.vim/pack/plugins/start/slimv/doc' +q

That is it! Slimv is set up. It's that straightforward. The
commands above show how to set up Slimv with just two shell
commands. You could also use a Vim plugin manager to install
Slimv for you but I am not going to cover that here.

This is an optional step. Slimv supports starting Swank server
automatically if you are running Vim in tmux, GNU Screen or a
desktop environment. To start tmux, enter this command:

tmux

If you use GNU screen or a desktop environment, you don't have
to run tmux.

If you do not use tmux, GNU Screen or a desktop environment,
then you must start Swank server manually as explained in the
next point.

This step is necessary only if you are not using tmux, GNU
Screen or a desktop environment. The following command shows
how to start Swank server manually:

sbcl --load ~/.vim/pack/plugins/start/slimv/slime/start-swank.lisp

If you are using tmux, GNU Screen or a desktop environment,
Slimv can start Swank server automatically when needed and you
don't need to perform this step.

Create a new Lisp source code file, say, foo.lisp
with this command:

vim foo.lisp

To connect to Swank server, enter the following command in
normal mode:

,c

If Vim is running within tmux, GNU Screen or desktop
environment, Slimv would automatically launch Swank server and
connect to it.

After Slimv connects to Swank successfully, Vim window should
split into two and the following prompt should appear in the new
split window:

CL-USER>

This is the integrated REPL. It is now alive and ready for
interactive programming.

We assume here that Slimv is using the default Slimv leader
key ,. If you have overridden the Vim leader key,
then the Slimv leader key might be same as the Vim leader key.
Enter the command :echo g:slimv_leader in Vim
command-line mode to find the leader key being used by Slimv.

Type some code into the buffer for the new file. To do so,
first type i to enter insert mode and type this code:

(format t "hello, world~%")

Type esc to return to normal mode.

To evaluate the current expression under the cursor, enter the
following command in normal mode:

,e

Both the current expression and its result should appear in the
REPL window.

The REPL is interactive. Type
ctrl+ww to switch to the REPL
window. Then type i to enter insert mode and type
this code:

(+ 1 2)

Type enter to evaluate the expression just like you
would do in a real REPL. The result should then appear in the
REPL.

Type esc to return to normal mode again. Use the
normal mode command
ctrl + w w to
switch between the split windows.

Now that you have got started with Slimv, here is a brief note
on uninstallation, in case you ever need it. If Slimv is
installed as described in point 2 above, enter the following
command to uninstall it:

rm -rf ~/.vim/pack/plugins/start/slimv

In steps 7 and 9, you may have noticed that as soon as you type an
opening parenthesis or double quotation mark, a matching closing one
is automatically inserted. That is done by the
Paredit plugin
which is bundled along with Slimv. Paredit ensures structured
editing of Lisp s-expressions and keeps all matched characters
(parentheses, brackets, braces, quotes) balanced. It also provides
many new keybindings to edit s-expressions conveniently. We will
look into Paredit in a little more detail in
the Get Started with
Paredit subsection later.

Get Started with Vlime and SBCL

Here are the steps to set up Vlime and use it:

Install the tools required to set up a Common Lisp development
environment with Vlime with this command:

sudo apt-get install vim sbcl git curl

Note that unlike Slimv, Vlime can work with the default Vim in
Debian, i.e. vim.basic. Vlime does not require Vim
with Python interface.

Install Quicklisp with these commands:

curl -O https://beta.quicklisp.org/quicklisp.lisp
sbcl --load quicklisp.lisp --eval '(quicklisp-quickstart:install)' --eval '(exit)'
sbcl --load ~/quicklisp/setup.lisp --eval '(ql:add-to-init-file)' --eval '(exit)'

Type enter in the end, when prompted, to complete the
installation.

Install Vlime and Paredit with these commands:

git clone https://github.com/l04m33/vlime.git ~/.vim/bundle/vlime
git clone https://github.com/kovisoft/paredit ~/.vim/pack/plugins/start/paredit
echo 'set runtimepath^=~/.vim/bundle/vlime/vim' >> ~/.vimrc
vim +'helptags ~/.vim/bundle/vlime/vim/doc' +'helptags ~/.vim/pack/plugins/start/paredit/doc' +q

Unlike Slimv, Vlime does not bundle
Paredit along
with itself. As explained in the previous section, it helps us
with structured editing of Lisp s-expressions.

I recommend that you install Paredit but in case you choose not
to, ensure that loading of filetype plugins is enabled by
entering the :filetype command in command-line
mode. The output should contain plugin:ON. If it
is off, add the command filetype plugin on
to ~/.vimrc to ensure that this is always on.
Vlime won't work without this being enabled. If you install
Paredit, you don't have to bother about this because Paredit
takes care of enabling this by default.

Create a new Lisp source code file, say, foo.lisp
with this command:

vim foo.lisp

To start Vlime server (a wrapper around Swank server) and
connect to it automatically, enter the following command in
normal mode:

\rr

We assume here that Vim <LocalLeader> is left
to its default, i.e. backslash. If it is mapped to some other
key combination, then that must be used instead of backslash in
the above command.

The first time this command is run after installing Vlime, it
installs Swank server using Quicklisp. Therefore, it can take a
while for Vlime server to start the first time this command is
run. On subsequent use of these commands, it would start faster
because it would be already installed.

The console output from Vlime server is displayed in a split
window. After Vlime successfully connects to Swank, the
following message is displayed at the bottom:

Vlime Connection 1 established.

After the above message appears, it is okay to close the split
window for Vlime server by entering this command in Vim
command-line mode:

:q

Vlime server would continue to run in background. The following
command can be used in normal mode to view the console output of
Vlime server anytime it is required:

\rv

Type some code into the buffer for the new file. To do so,
first type i to enter insert mode and type this code:

(format t "hello, world~%")

Type esc to return to normal mode.

To evaluate, the current expression under the cursor, enter the
following command in normal mode:

\ss

Both the current expression and its result should appear in the
REPL window.

Unlike Slimv, the REPL window of Vlime is not
interactive. Its nomodifiable option is set, so we
cannot type code directly into the REPL window. This can be a
bit of a problem if we want to type arbitrary expressions into
the REPL and execute them. To mitigate this shortcoming to some
extent, Vlime provides an alternative way to evaluate the
current expression known as the interaction mode. This is
explained in the next point.

Enable interaction mode by entering this command in normal mode:

\i

The same command disables interaction mode, i.e. this command
toggles the state of interaction mode between on and off. When
interaction mode is on, evaluate an expression under the cursor
by simply pressing enter in normal mode.

Now that you have got started with Vlime, here is a brief note
on uninstallation, in case you ever need it. If Quicklisp and
Vlime are installed as described in the points 2 and 3 above,
run these commands to uninstall them:

rm -rf ~/quicklisp ~/.vim/bundle/vlime ~/.vim/pack/plugins/start/paredit
sed -i.bkp '/runtimepath.*vlime/d' ~/.vimrc

Optionally, remove ~/.sblrc or edit it to remove
the code pertaining to loading
quicklisp/setup.lisp.

Get Started with Paredit

You have already got started with Paredit when you wrote Lisp code
while following one of the previous two subsections. The moment you
typed an opening parenthesis, Paredit inserted a closing one for you
automatically. Paredit keeps all matched characters such as
parentheses, double quotes, etc. balanced when you edit code. Here
is a very brief exercise to quickly get started with some of the
very basic features of Paredit:

Create a new Lisp source code file, say, foo.lisp
with this command:

vim foo.lisp

Type i to enter insert mode and then type only this:

(defun square (x

At this point, Paredit should have inserted the two closing
parentheses automatically. The code should look like this:

(defun square (x))

The cursor should be situated just after the parameter
x. The block above shows where the cursor should
be.

While you are still in insert mode, type the first closing
parenthesis. Yes, type it even if the closing parenthesis is
already present. The cursor should now skip over the first
closing parenthesis like this:

(defun square (x))

Of course, there was no need to type the closing parenthesis
because it was already present but typing it out to skip over it
is more efficient than escaping to normal mode, then moving over
it and then entering insert mode again. This is, in fact, a
very nifty feature of Paredit. We can enter code with the same
keystrokes as we would without Paredit.

You should still be in insert mode. Type enter to
create a new line below. Now one of two things is going to
happen. If electric return is disabled, then a newline is
inserted as expected like this:

(defun square (x))
)

If electric return is enabled, two newlines are inserted to
create an empty line in between:

(defun square (x)

)

In both cases, indentation of two spaces is inserted
automatically. The new empty line inserted by electric return
allows linewise editing of the code to be entered in this empty
line.

The electric return feature is enabled by default in both
Paredit and Slimv. It works by Paredit remapping the "enter"
key (<CR>) in insert mode to a function that
inserts electric return. Slimv needs to remap the "enter" key
to present the argument list of the current function but it
takes care of performing an electric return before showing the
argument list. Vlime, however, forgets to perform electric
return before showing the argument list, so this feature does
not work in Vlime.

For now, we will continue with the assumption that electric
return is enabled and working fine. If it is disabled or if it
is not working for you, ignore the steps that discuss electric
return.

Now, type only this:

(* x x

Again, Paredit would have inserted the closing parenthesis
automatically. The code should look like this now:

(defun square (x)
(* x x)
)

Now, type one more closing parenthesis to advance past the
automatically inserted closing parenthesis like this:

(defun square (x)
(* x x)
)

Then type another closing parenthesis. Paredit would now pick
the lone closing parenthesis that is present in its own line and
move it at the end of the current line like this:

(defun square (x)
(* x x))

This behaviour of consuming the extra newline inserted by an
electric return on typing a closing parenthesis helps the code
to conform to the popular Lisp coding convention of putting all
the consecutive closing parentheses next to each other in the
same line. In other words, typing closing parentheses
re-gathers electric returns when applicable.

Let us see what happens if we try to delete the opening
parenthesis around the product function (the *
function). Type esc to return to normal mode. Then
enter h in normal mode to move the cursor one place
left so that the cursor is placed on the parenthesis just after
the last x in the code like this:

(defun square (x)
(* x x))

Type x to delete the closing parenthesis the cursor
is on. Nothing gets deleted! Instead the cursor just skips
over the parenthesis like this:

(defun square (x)
(* x x))

Paredit refuses to delete the closing parenthesis because it
encloses a non-empty list. It would have deleted the closing
parenthesis along with the opening one if the list were empty.
This is Paredit trying to ensure that the s-expressions remain
valid while editing.

Note that in this step, h is a regular Vim motion
command. In Vim, by default, x deletes the character
under the cursor, but when Paredit is enabled, it remaps this
command to behave the way it did in this step to ensure that the
parentheses remain balanced.

Let us now try to delete the current line. Type
dd to do so. The result looks like this:

(defun square (x)
)

Note how the closing parenthesis has been left intact to keep
the parentheses balanced. Again, Paredit has remapped the
dd command to produce this behaviour.

Now type da( to delete the
entire defun expression. The buffer should look
empty now.

Type i to enter insert mode and type out the
following code:

(list (* 10 20) (+ 30 40))

Type esc to return to normal mode.
Type h(hh to place
the cursor on the closing parenthesis of the first expression.

(list (* 10 20) (+ 30 40))

Now type ,>. The closing parenthesis
of the first expression moves right to slurp the next
expression. The buffer looks like this:

(list (* 10 20 (+ 30 40)))

We assume here that Paredit is using the default Paredit leader
key ,. If you have overridden the Vim leader key,
then the Paredit leader key might be same as the Vim leader key.
Enter the command :echo g:paredit_leader in Vim
command-line mode to find the leader key being used by Paredit.

Now type ,<. The closing parenthesis
of the outer expression moves left to barf the inner
expression out. The buffer looks like this again:

(list (* 10 20) (+ 30 40))

While the cursor is on a parenthesis, the normal mode
commands ,< or ,> can be used in
this manner to move the parenthesis left or right respectively,
thereby slurping or barfing expressions.

That was a very brief overview of what Paredit can do. There is a
lot more to Paredit than what is described above. Paredit has a
rich set of keybindings to make editing s-expressions very
convenient. Enter :help paredit-keys in command-line
mode to see the list of the keybindings.

I think it is a good idea to read the entire Paredit documentation.
Enter :help paredit to do so. It is about 500 lines
long and takes about 30 to 40 minutes to read. The time spent
reading this documentation is worth it because it makes editing Lisp
code very pleasant and productive.

Use Debugger and Inspector

After getting started with Slimv or Vlime, the very next thing you
might want to know is how to work with the debugger. The debugger
window comes up whenever an error or an unhandled condition occurs.
It might look quite scary to a beginner, so it is a good idea to
become comfortable with it as soon as possible. Fortunately, both
Slimv and Vlime provide excellent key-bindings to inspect the error
or dismiss it to return to the source code buffer quickly and
easily.

Use Debugger and Inspector with Slimv

The following steps trigger an error and then show how to work with
the debugger in Slimv:

Create a file with Vim, say foo.lisp and enter the
following code into it:

(defun square (x)
(* x x))

(square "foo")

Enter ,b to evaluate the buffer.

As soon as the defective form (square "foo") gets
evaluated, an error occurs. The error, possible restarts and
the backtrace is displayed in a new split window for SLDB. SLDB
stands for Slime Debugger. Here is an example of what may
appear in the SLDB window:

The value
"foo"
is not of type
NUMBER
when binding SB-KERNEL::X
[Condition of type TYPE-ERROR]

Restarts:
0: [RETRY] Retry SLIME REPL evaluation request.
1: [*ABORT] Return to SLIME's top level.
2: [ABORT] abort thread (#<THREAD "repl-thread" RUNNING {1003274E23}>)

Backtrace:
0: (SB-KERNEL:TWO-ARG-* "foo" "foo") [external]
1: (SQUARE "foo")
2: (SB-DEBUG::TRACE-CALL #<SB-DEBUG::TRACE-INFO SQUARE> #<FUNCTION SQUARE> "foo")
3: (SB-INT:SIMPLE-EVAL-IN-LEXENV (SQUARE "foo") #<NULL-LEXENV>)
4: (EVAL (SQUARE "foo"))
5: (SWANK::EVAL-REGION "(defun square (x) ..)
6: ((LAMBDA NIL :IN SWANK-REPL::REPL-EVAL))
...

The ellipsis in the end is added by me to denote that the actual
output has been truncated in this article for the sake of
brevity.

In the SLDB window, move the cursor to the second line of
backtrace, i.e. on the following line:

1: (SQUARE "foo")

Then type enter. This line should now unfold to show
the following details:

1: (SQUARE "foo")
in "(SB-INT:NAMED-LAMBDA SQUARE-----------------------------------
Locals:
X = "foo"

Move the cursor to the line that begins with in,
i.e. on this line:

in "(SB-INT:NAMED-LAMBDA SQUARE-----------------------------------

Then type enter. Some information about its source
code should appear like this:

1: (SQUARE "foo")
in "(SB-INT:NAMED-LAMBDA SQUARE
(X)
(BLOCK SQUARE (#:***HERE*** (* X X))))" byte 1
Locals:
X = "foo"

Move the cursor to the following line:

X = "foo"

Then type ,i to inspect this variable. A
prompt would appear to confirm the variable name. Type
enter to confirm. An inspector window should now
appear with more details about this variable. This window should
look like this:

Inspecting #<(SIMPLE-ARRAY CHARACTER (3)) {100478AFAF}>
--------------------
Press <F1> for Help

Dimensions: (3)
Element type: CHARACTER
Total size: 3
Adjustable: NIL
Fill pointer: NIL
Contents:
0: #\f
1: #\o
2: #\o

[<<] Exit Inspector

Type enter to inspect any object under the cursor and
drill down further.

Type backspace in normal mode to return to the
previous object.

Enter ,q in normal mode to quit the
inspector.

Finally, move the cursor to the following line in the SLDB
window:

1: [*ABORT] Return to SLIME's top level.

Then type enter to execute this restart.
Alternatively, enter ,a in normal mode to
select the abort restart and quit to the previous level or
,q to quit to top level.

Most of the times when an error occurs, I quickly take a look at the
stack trace to realise that I have made a silly mistake and enter
the ,q command to abort and quit to top level.
This can be quite convenient because it allows returning from
debugging to coding very quickly with only two keystrokes.

Use Debugger and Inspector with Vlime

The following steps trigger an error and then show how to work with
the debugger in Vlime:

Create a file with Vim, say foo.lisp and enter the
following code into it:

(defun square (x)
(* x x))

(square "foo")

Save the file, connect to Vlime server and enter
\of in normal mode to compile
the entire buffer.

As soon as the defective form (square "foo") gets
evaluated, an error occurs. The error, possible restarts and
the backtrace is displayed in a new split window for SLDB. SLDB
stands for Slime Debugger. Here is an example of what may
appear in the SLDB window:

Thread: 1; Level: 1

The value
"foo"
is not of type
NUMBER
when binding SB-KERNEL::X
[Condition of type TYPE-ERROR]

Restarts:
0. *ABORT - Return to SLIME's top level.
1. ABORT - abort thread (#<THREAD "worker" RUNNING {10045D6F83}>)

Frames:
0. (SB-KERNEL:TWO-ARG-* "foo" "foo") [external]
1. (SQUARE "foo")
2. (SB-FASL::LOAD-FASL-GROUP #S(SB-FASL::FASL-INPUT :STREAM #<SB-SYS:FD-STREAM for "file /home/susam/foo.fasl" {10045E76A3}> :TABLE #(41 #<PACKAGE "SB-IMPL"> SB-IMPL::%DEFUN #<PACKAGE "COMMON-LISP-USER">..
3. (SB-FASL::LOAD-AS-FASL #<SB-SYS:FD-STREAM for "file /home/susam/foo.fasl" {10045E76A3}> NIL NIL)
4. ((FLET SB-FASL::THUNK :IN LOAD))
5. (SB-FASL::CALL-WITH-LOAD-BINDINGS #<CLOSURE (FLET SB-FASL::THUNK :IN LOAD) {7F7B9B0B60BB}> #<SB-SYS:FD-STREAM for "file /home/susam/foo.fasl" {10045E76A3}>)
...

The ellipsis in the end is added by me to denote that the actual
output has been truncated in this article for the sake of
brevity.

In the SLDB window, move the cursor to the second line of
backtrace, i.e. on the following line:

1. (SQUARE "foo")

Then type d. A new split window should appear with
the following details about this frame:

Frame: 1 (Restartable)

Locals:
X: "foo"

Location:
File: /home/susam/foo.lisp
Position: 20
Snippet:
(* x x))

(square "foo")

While the cursor is on the same line as mentioned in the
previous point, type i to bring up the inspector
window for this frame.

In the inspector window, type i to enter insert mode.
Enter the following variable name in insert mode:

x

Then type esc to return to normal mode. Then type
enter. The following details about the variable
x should now appear in the inspector window:

#<(SIMPLE-ARRAY CHARACTER (3)) {1004617ABF}>
============================================

Dimensions: (3)
Element type: CHARACTER
Total size: 3
Adjustable: NIL
Fill pointer: NIL
Contents:
0: #\f
1: #\o
2: #\o

Type enter to inspect any object under the cursor and
drill down further.

Type p to return to the previous object.

Enter the regular Vim command :q in command-line
mode to quit the inspector window.

Finally, move the cursor to the following line in the SLDB
window:

1: [*ABORT] Return to SLIME's top level.

Then type enter to execute this restart.
Alternatively, we can enter a in normal mode to
select the abort restart to return to the previous level. At
this time, there is no command to return to SLIME's top level.

Trace Function
Trace Function in Slimv

The following steps show how to get started with tracing functions
in Slimv:

Create a file with Vim, say foo.lisp and enter the
following code into it:

(defun square (x)
(* x x))

(square (square 2))

Enter ,b in normal mode to evaluate the
entire buffer.

Place the cursor on the function name, i.e. on
square and enter ,t in normal
mode to toggle tracing for this function. A prompt appears to
confirm the function name. Type enter to confirm.

While the cursor is on the last expression, enter
,d in normal mode to evaluate the top-level
form. The following output appears in the REPL buffer.

(square (square 2))
0: (SQUARE 2)
0: SQUARE returned 4
0: (SQUARE 4)
0: SQUARE returned 16
16

This output contains information about each call to the traced
function, arguments passed to it and the return values.

Trace Function in Vlime

It takes a little more work to start tracing functions in Vlime.
The following steps show how to do it:

Add the following statement to ~/.vimrc:

let g:vlime_contribs = ['SWANK-ASDF', 'SWANK-PACKAGE-FU',
\ 'SWANK-PRESENTATIONS', 'SWANK-FANCY-INSPECTOR',
\ 'SWANK-C-P-C', 'SWANK-ARGLISTS', 'SWANK-REPL',
\ 'SWANK-FUZZY', 'SWANK-TRACE-DIALOG']

The above variable defines the list of Swank contrib modules to
load while initialising a Vlime connection. All modules
mentioned above except the last one are loaded by default. The
SWANK-TRACE-DIALOG module is not loaded by default
but this module is necessary for tracing functions, so in order
to load it, we define this variable to load this module in
addition to all the other modules that are loaded by default.

Create a file with Vim, say foo.lisp and enter the
following code into it:

(defun square (x)
(* x x))

(square (square 2))

Save the file, connect to Vlime server and enter
\of in normal mode to compile
the entire buffer.

Enter \TD in normal mode to
show the trace dialog in a split window.

Enter ctrl + w w
in normal mode to go back to the source code window.

Place the cursor on the function name, i.e. on
square and enter
\TT in normal mode to toggle
tracing for this function.

While the cursor is on the last expression, enter
\st in normal mode to evaluate
the top-level form.

Enter ctrl + w w
in normal mode twice to go to the trace window.

Under Trace Entries, place the cursor on
[refresh] and type enter.

Then place the cursor on [fetch next batch] and
type enter. Two results should appear for the two
square calls that were made due to step 7. The trace
information would be folded under each call.

Move the cursor to each fold line and enter
zo in normal mode to open the fold. After
opening both the folds, the following result should be visible:

0 - COMMON-LISP-USER::SQUARE
> 2
< 4
1 - COMMON-LISP-USER::SQUARE
> 4
< 16
16

The lines starting with > show the arguments and
the ones starting with < show the return values.

Nifty Features

In this section, we will go over some of the nifty features that
these plugins offer. Not all features will be covered here. I have
chosen only a few features for the discussion here that I felt would
be useful to beginners and at the same time also demonstrate the
versatility of these plugins.

Evaluate Top-Level Form

In the previous sections, we saw how to evaluate the current
expression under the cursor. In this section, we will see how to
evaluate the top-level expression around the current cursor
position. Let us do a small exercise to see this:

Create a file with Vim, say foo.lisp and enter the
following code into it:

(+ 1 (* 2 (/ 6 2)))

With Slimv or Vlime connected to Swank, let us do a quick recap
of how to evaluate the current expression.

With Slimv, enter the normal mode
command ,e to evaluate the current
expression.

With Vlime, enter the normal mode command
\ss to evaluate the current
expression.

The current expression, i.e. (/ 6 2) should get
evaluated and the result 3 should appear in the
REPL buffer.

Let us now see how to evaluate the top-level expression.

With Slimv, enter the normal mode
command ,d to evaluate the top-level
expression.

With Vlime, enter the normal mode command
\st to evaluate the top-level
expression.

The top-level expression should get evaluated and the result
7 should appear in the REPL buffer.

Rainbow Parentheses

Rainbow parentheses make it easy to see matching parentheses by
colouring different levels of parentheses with different colours.
Matching parentheses have the same colour. To enable this feature
in Slimv, add this command to ~/.vimrc:

let g:lisp_rainbow=1

This feature is not available in Vlime. But there are several Vim
plugins that support rainbow parentheses. Here are the steps to
install one such plugin that is quite popular:

git clone https://github.com/junegunn/rainbow_parentheses.vim.git ~/.vim/pack/plugins/start/rainbow_parentheses
echo 'autocmd FileType lisp,scheme,clojure RainbowParentheses' >> ~/.vimrc

In case you ever want to uninstall it, enter these commands:

rm -rf ~/.vim/pack/plugins/start/rainbow_parentheses
sed -i.bkp '/autocmd.*RainbowParentheses/d' ~/.vimrc
Argument List

You must have seen this feature already while trying out the
sections earlier. While editing a Lisp source file, after typing a
function name, as soon as a space is typed or the enter key is
typed, the argument list for the function appears to serve as a
reference. In Slimv, the argument list appears in the status line
at the bottom. In Vlime, the argument list appears in a split
window at the top.

Omni-Completion

Type a function name partially, e.g. form and type
tab while still in insert mode. The omni-completion menu
should appear with the list of completions if there are multiple
choices. Type ctrl+n to select the next choice
and ctrl+p to select the previous choice.
Selecting a choice also immediately inserts that choice in the
buffer. This works in both Slimv and Vlime. In Slimv, we can also
type tab to select the next choice.

By default, omni-completion is fuzzy. For example,
type wl and type tab and omni-complete
should insert write-line automatically as well as show
other matching choices.

Describe Symbol

With Slimv, enter the normal mode command ,s
to describe the symbol under the cursor. This brings up the
documentation of the symbol in the Vim message area. This feature
works while editing Common Lisp and Clojure source files but not
while editing Scheme source file. This feature is not supported for
Scheme at this time. See
the Other Lisp Dialects
section for details on how to set up Slimv with Clojure and MIT/GNU
Scheme.

With Vlime, enter \da in normal
mode to describe the symbol under the cursor. This brings up the
documentation of the symbol in a split window.

Expand Macro

Here is an excercise that shows how to expand macros interactively
while editing a Lisp source file:

Create a file with Vim, say foo.lisp and enter the
following code into it:

(defmacro calc (a op b)
(list op a b))

(defmacro square (x)
(list 'calc x '* x))

(square 2)

With Slimv, enter ,b in normal mode to
evaluate the entire buffer.

With Vlime, save the file, connect to Vlime server and type
\of in normal mode to compile the
entire buffer.

With Slimv, while the cursor is on the last expression, enter
,1 in normal mode to expand the macro form
once.

With Vlime, enter \m1 in normal
mode to do the same thing.

The following expansion should appear as the result:

(CALC 2 * 2)

Slimv displays the expansion in the REPL buffer whereas Vlime
displays it in a new split window.

With Slimv, enter ,m in normal mode to
recursively expand the current expression until it is no longer
a macro.

With Vlime, enter \ma in normal
mode to do the same thing.

The following expansion should appear as the result:

(* 2 2)

Cross Reference

Here is an exercise that shows how to use the cross-reference
commands in Slimv and Vlime:

Create a file with Vim, say foo.lisp and enter the
following code into it:

(defun square (x)
(* x x))

(defun square-of-sum (x y)
(square (+ x y)))

(defun sum-of-squares (x y)
(+ (square x) (square y)))

(square-of-sum 2 3)
(sum-of-squares 2 3)

With Slimv, enter ,b in normal mode to
evaluate the entire buffer.

With Vlime, save the file, connect to Vlime server and type
\of in normal mode to compile the
entire buffer.

With Slimv, place the cursor on any occurrence of the symbol
square and enter
,xl in normal mode. A prompt
would appear to confirm the symbol name. Type enter
to confirm. The list of all callers should now appear in the
REPL buffer.

With Vlime, place the cursor on any occurrence of the symbol
square and enter \xc
in normal mode to list all callers of the function. The output
appears in a split window containing the cross reference (xref)
buffer. Type enter on any item in the xref buffer and
Vlime will take you directly to the referenced location.

Other Common Lisp Implementations

The previous sections used SBCL as the implementation of Common
Lisp. How well do Slimv and Vlime work with other Common Lisp
implementations?

I have found that both plugins are pretty well tested with SBCL.
However, they may not be so well tested with other implementations.
Due to the lack of sufficient testing with Common Lisp
implementations other than SBCL, certain errors may occur while
using other implementations. Sometimes it is possible to work
around these errors and sometimes it isn't. We will see an example
of this in an upcoming section when we try to start Swank server
automatically using Vlime and CLISP.

For this section, I choose CLISP and Embeddable Common-Lisp (ECL) as
two other implementations of Common Lisp that will be used with
Slimv and Vlime. After following the upcoming subsections, you
should get the hang of how to make Slimv or Vlime work with other
implementations of Common Lisp.

Use Slimv with CLISP

If you have read and tried the steps in the
Get Started with
Slimv and SBCL section, it is going to be quite easy to use
Slimv with CLISP. The steps are similar with a few minor
modifications. They are explained below:

Uninstall SBCL and install CLISP with these commands:

sudo apt-get remove sbcl
sudo apt-get install clisp

To start Swank server manually, enter this command:

clisp ~/.vim/pack/plugins/start/slimv/slime/start-swank.lisp

Then edit a Lisp source file and enter the normal command
,c to connect to it and bring up the REPL
window.

To start Swank automatically from Slimv, there is nothing more
to be done. Just edit a Lisp source file and enter the normal
mode command ,c. While running in GNU
Screen, tmux or a desktop environment, Slimv can automatically
detect CLISP and start Swank server with it.

In general, to start Swank server manually with another Common Lisp
implementation, we need to figure out how to load
start-swank.lisp with it.

Use Slimv with ECL

The steps to use Slimv with Embeddable Common-Lisp (ECL) are very
similar too. Once again, only if we need to start Swank server
manually, we need to figure out the command to do so. Otherwise,
there is no other difference. Here are the steps:

Ensure that SBCL and CLISP are uninstalled and ECL is installed.

sudo apt-get remove sbcl clisp
sudo apt-get install ecl

To start Swank server manually, enter this command:

ecl --load ~/.vim/pack/plugins/start/slimv/slime/start-swank.lisp

Then edit a Lisp source file and enter the normal command
,c to connect to it and bring up the REPL
window.

To start Swank automatically from Slimv, there is nothing more
to be done. Just edit a Lisp source file and enter the normal
mode command ,c. While running in GNU
Screen, tmux or a desktop environment, Slimv can automatically
detect CLISP and start Swank server with it.

There is a possible timeout issue to be aware of though. ECL
can take a minute or two to compile the code it loads the first
time Swank server is started. However, Slimv has a default
timeout period of 20 seconds, so Slimv may fail with the
following error message:

SWANK server is not running. Press ENTER to continue.

If this happens, just wait for ECL to complete compiling Swank
server. Once it starts Swank server, enter the normal mode
command ,c again and it should connect
immediately.

Use Vlime with CLISP

This subsection assumes that you have already read and tried the
Get Started with
Vlime and SBCL section, so you are familiar with Vlime
basics. Now we will see what more it takes to use Vlime with CLISP
in the steps below:

Let us assume we want to start afresh with CLISP, i.e. we do not
have previous artefacts created by SBCL. To clean up old
artefacts, enter these commands:

rm -rf ~/.sbclrc ~/quicklisp
sudo apt-get remove sbcl

Install CLISP with this command:

sudo apt-get install clisp

Install Quicklisp using CLISP with these commands:

curl -O https://beta.quicklisp.org/quicklisp.lisp
clisp -i quicklisp.lisp -x '(quicklisp-quickstart:install)'
clisp -i ~/quicklisp/setup.lisp -x '(ql:add-to-init-file)'

Type enter in the end, when prompted, to complete the
installation.

Add the following code to ~/.vimrc:

let g:vlime_cl_impl = 'clisp'
function! VlimeBuildServerCommandFor_clisp(vlime_loader, vlime_eval)
return ['clisp', '-i', a:vlime_loader,
\ '-x', a:vlime_eval,
\ '-repl']
endfunction

Unlike Slimv, automatic start of Swank server with Common Lisp
implementations other than SBCL are not supported out of the
box, so the above Vim script tells Vlime how to start Swank
server with CLISP. The -repl option is used to
work around an issue that is explained in the next point.

Vlime is now ready to be used with CLISP. Just edit a Lisp
source file and enter the normal mode command
\rr to start Swank server and
connect to it automatically.

You may see the following error in the SLIME debugger
(sldb) split window:

SOCKET-STATUS on #1=#<INPUT STRING-INPUT-STREAM> is illegal

Despite the above error, the following message should appear at
the bottom:

Vlime Connection 1 established.

If the above message occurs, you can ignore this error, close
the debugger window as well as the console output window and
continue to use Vlime normally.

The -repl option used in the previous step ensures
that the REPL starts despite this error. Without it, this step
would not have succeeded. This is what I meant when I said
earlier that we may need to work around certain errors while
using these plugins with a Common Lisp implementation other than
SBCL.

Use Vlime with ECL

Here are the steps to use Vlime with ECL:

Let us assume we want to start afresh with ECL, i.e. we do not
have previous artefacts created by SBCL or ECL. To clean up old
artefacts, enter these commands:

rm -rf ~/.sbclrc ~/.clisprc.lisp ~/quicklisp
sudo apt-get remove sbcl clisp

Install ECL with this command:

sudo apt-get install ecl

Install Quicklisp using ECL with these commands:

curl -O https://beta.quicklisp.org/quicklisp.lisp
ecl --load quicklisp.lisp --eval '(quicklisp-quickstart:install)' --eval '(quit)'
ecl --load ~/quicklisp/setup.lisp --eval '(ql:add-to-init-file)' --eval '(quit)'

Type enter in the end, when prompted, to complete the
installation.

Add the following code to ~/.vimrc:

let g:vlime_cl_impl = 'ecl'
function! VlimeBuildServerCommandFor_ecl(vlime_loader, vlime_eval)
return ['ecl', '--load', a:vlime_loader,
\ '--eval', a:vlime_eval]
endfunction

Edit a Lisp source file and enter the normal mode command
\rr to start Swank server and
connect to it automatically.

Other Lisp Dialects

So far, we have seen how to use Slimv or Vlime with a Common Lisp
implementation. Now let us see how well these plugins work with
other Lisp dialects. Vlime does not support other Lisp dialects.
It supports Common Lisp only. Slimv supports two other popular
dialects of Lisp: Scheme and Clojure. In the next two subsections,
we see how

Use Slimv with MIT/GNU Scheme

Slimv is documented to work with MIT/GNU Scheme on Linux only.
Enter
:help slimv-installation in Vim to read more about it.
It says the following under the "Prerequisites" section.

Lisp or Clojure or MIT/GNU Scheme (Linux only) installed.

Further, the Swank loader script for MIT/GNU Scheme named
swank-mit-scheme.scm says the following in its source code
comments:

You need MIT/GNU Scheme 9.2

At the time of writing this article, I have confirmed that both
these requirements indeed need to be met to use Slimv with MIT/GNU
Scheme. Here are the steps to use Slimv with MIT/GNU Scheme:

Install MIT/GNU Scheme with this command:

sudo apt-get mit-scheme

Ensure that vim-nox, tmux and Slimv are installed
as explained in the
Get Started with
Slimv and SBCL subsection earlier.

This is an optional step. To start Swank server automatically
from Slimv, run Vim in tmux, GNU Screen or a desktop
environment. In this article, we use tmux, so start tmux with
this command:

tmux

This step is necessary only if you are not using tmux, GNU
Screen or a desktop environment. In such a case, enter this
command to start Swank server manually:

scheme --load ~/.vim/pack/plugins/start/slimv/slime/contrib/swank-mit-scheme.scm

Create a new Scheme source code file, say, foo.scm
with this command:

vim foo.scm

To connect to Swank server, enter the following command in Vim
normal mode:

,c

Type some code into the buffer for the new file. To do so,
first type i to enter insert mode and type this code:

(display "hello, world\n")

Type esc to return to normal mode. To evaluate, the
current expression under the cursor, enter the following command
in normal mode:

,e

Both the current expression and its result should appear in the
REPL window.

I have confirmed that the steps above work fine with MIT/GNU Scheme
9.1.1 on Debian GNU/Linux 9.11 (stretch). Like I mentioned before,
Slimv requires Linux to work with MIT/GNU Scheme. For example,
trying to start Swank server with MIT/GNU Scheme 9.2 on macOS High
Sierra 10.13.6 fails with this error:

; /usr/local/Cellar/mit-scheme/9.2_2/lib/mit-scheme-c/include/config.h:879:10:
fatal error: 'sys/types.h' file not found

Further, the version of MIT/GNU Scheme really needs to be 9.x. For
example, when I try to start Swank with MIT/GNU Scheme 10.1.5 on
Debian GNU/Linux 10.1 (buster), the following error occurs:

;The object #[package 12 (user)], passed as an argument to ->environment, is not an environment.
Use Slimv with Clojure

Slimv works fine with Clojure too. However, it may have some
trouble locating Clojure on the system if we attempt to start Swank
server automatically with Clojure. That is because where and how
Clojure is installed varies from operating system to operating
system and also depends on the installation procedure chosen to set
up Clojure.

On Unix-like systems, Slimv looks for JAR files that match the glob
pattern clojure*.jar at paths that match the glob
pattern /usr/local/bin/*clojure*
and ~/*clojure, in that order. On Windows, it looks
for the JAR files at directory paths that match the glob
pattern C:\*clojure* and C:\*clojure*\lib.
Additionally, Slimv also looks for the JAR files at the paths
mentioned in the PATH environment variable. There are
a few more strategies too to locate Clojure but we will not get into
that here.

In this section, I will show how to build Clojure from source with
Maven and install it at ~/clojure/clojure.jar, a path
Slimv can easily find, so installing it here would mean that the
steps below would work everywhere regardless of the operating
system. If you are on Windows, install Clojure
at C:\clojure\clojure.jar instead.

Here are the steps to install Clojure at
~/clojure/clojure.jar and use it with Slimv:

Choose one of the two sets of commands below to install Maven:

# On Debian, Ubuntu, etc.
sudo apt-get install maven

# On macOS
brew install openjdk maven
export JAVA_HOME=/usr/local/opt/openjdk
export PATH="$JAVA_HOME/bin:$PATH"

Enter these commands to install Clojure:

git clone https://github.com/clojure/clojure.git ~/clojure
git -C ~/clojure checkout clojure-1.10.1
mvn -f ~/clojure/pom.xml -Plocal -Dmaven.test.skip=true package

Ensure that vim-nox, tmux and Slimv are installed
as explained in
the Get Started
with Slimv and SBCL subsection earlier.

This is an optional step. To start Swank server automatically
from Slimv, run Vim in tmux, GNU Screen or a desktop
environment. In this article, we use tmux, so start tmux with
this command:

tmux

This step is necessary only if you are not using tmux, GNU
Screen or a desktop environment. In such a case, enter these
commands to start Swank server manually:

SWANK_DIR=~/.vim/pack/plugins/start/slimv/swank-clojure
java -cp "$HOME/clojure/clojure.jar:$SWANK_DIR" clojure.main -i "$SWANK_DIR/swank/swank.clj" -e '(swank.swank/start-repl)' -r

Create a new Clojure source code file, say, foo.clj
with this command:

vim foo.clj

To connect to Swank server, enter the following command in Vim
normal mode:

,c

Type some code into the buffer for the new file. To do so,
first type i to enter insert mode and type this code:

(println "hello, world")

Type esc to return to normal mode. To evaluate, the
current expression under the cursor, enter the following command
in normal mode:

,e

Both the current expression and its result should appear in the
REPL window.

Comparison of Slimv and Vlime

Finally, let me provide a comparison of both Slimv and Vlime side by
side. This comparison table below is not exhaustive. There are
more differences between the tools than what is mentioned below.

Slimv
Vlime

Slimv's directory structure conforms to the directory
structure of plugins in a Vim package as well as the default
directory structure expected by popular Vim plugin managers,
so installing Slimv is quite straightforward.

Vlime's directory structure does not conform to the
directory structure of plugins in a Vim package or the
default directory structure expected by popular Vim plugin
managers. As a result, Vim's native support for packages
cannot be used to install Vlime. Installing it via a plugin
manager requires fiddling with
Vim's runtimepath option in order to load it
successfully.

Slimv requires a Vim package that is compiled with support
for Python interface.

Vlime does not have this requirement. It can work with
basic Vim that does not have Python interface.

Slimv requires Vim to be running within tmux, GNU Screen or
a desktop environment to be able to start Swank server
automatically. If you have none of these, Swank server
needs to be started manually.

Vlime does not require tmux, GNU Screen or a desktop
environment in order to start Vlime server automatically.
It can start Vlime server on its own.

Slimv does not require Quicklisp to install Swank. Slimv
bundles the Swank server code with itself.

Vlime requires Quicklisp to be installed. It relies on
Quicklisp to install Swank the first time it is needed.

Slimv bundles Paredit with itself. Installing Slimv also
provides Paredit.

Vlime does not bundle Paredit with itself. Paredit needs to
be installed separately.

The buffer for REPL is interactive in Slimv. We can type
code directly into the REPL window and type enter
to execute it.

The buffer for REPL is not interactive in Vlime. Its
nomodifiable option is set, so we cannot type
code directly into the REPL window.

Paredit electric returns work fine with Slimv. Slimv remaps
the "enter" key to show argument list of the current
function after inserting electric returns. It takes care of
preserving the electric return functionality of Paredit.

Paredit electric returns do not work fine with Vlime. Vlime
remaps the "enter" key to show argument list of the current
function without inserting electric returns.

Slimv shows argument list of a function, symbol description,
etc. in the status line or message area at the bottom.

Vlime shows argument list of a function, symbol description,
etc. in separate split windows. These are extra windows to
skip over while cycling between windows with the normal mode
ctrl + w w
command which could feel inconvenient.

Slimv shows trace results, macro expansion, cross reference,
etc. in the REPL buffer.

Vlime shows resultions of trace results, macro expansion,
cross reference, etc. in split windows. These are extra
windows to skip over while cycling between windows. This
could feel inconvenient.

Slimv supports programming in Common Lisp, MIT/GNU Scheme
and Clojure.

Vlime supports programming in Common Lisp only. It does not
support Scheme or Clojure.

Slimv supports rainbow parentheses by adding
let g:lisp_rainbow=1 to ~/.vimrc.
With this feature, parentheses at different levels have
different colours and matching parentheses have the same
colour.

Vlime does not have rainbow parentheses. However, this is
not a major problem because there are several independent
plugins available that provide rainbow parentheses.

Slimv cross-reference commands do not help us to jump
directly to a function listed in the results.

Vlime cross-reference commands create a xref buffer that
allows us to jump directly to a function listed in the
results by moving the cursor to the function name in the
xref buffer and typing enter.

Quick Recommendation

If you are looking for a quick recommendation on which plugin to
use, I am going to recommend Slimv. It has been around for much
longer. It supports a wider variety of Lisp implementations. I
find its default key bindings more convenient. A truly interactive
REPL buffer is also a bonus. Also, Slimv supports Scheme and
Clojure whereas Vlime does not. Having said that, I think it is a
good idea to try out both the plugins on your own and then find out
which one suits you more.

Disclosure

Four bugs were harmed while writing this article!

While writing this article, I found the following four bugs in Slimv
which were then promptly squashed:
#87,
#88,
#89 and
#90.

References

Couple of Emacs hacks

Vim 8.0 Released

Slimv Tutorial

A Tutorial for Vlime

Comments |
#lisp |
#programming |
#vim |
#technology |
#how-to

Home
Links
Feed
Subscribe
Wander
About
Codeberg
Mastodon

© 2001–2026 Susam Pal

The integration of Lisp into the Vim editor has evolved over time, with Slimv and Vlime serving as prominent plugins offering interactive programming environments. The background context for these tools stems from Lisp, a programming language characterized by its prefix notation and powerful macro system, which has historically inspired profound intellectual engagement among computer scientists. The foundation for these Vim integrations draws from the concept of SLIME, which is an Emacs mode providing Read-Eval-Print-Loop (REPL) capabilities, which operated via a client-server architecture using a Swank backend server written in Common Lisp.

Slimv, developed by Tamas Kovacs, was released in 2009, and Vlime, created by Kay Z, followed in 2017, both aiming to replicate this interactive Lisp environment within the Vim framework. Both plugins utilize a client-server architecture, relying on the Swank server for runtime execution, although they handle the server integration differently. Slimv bundles a modified version of the Swank server, whereas Vlime provides a wrapper that automatically downloads and manages the Swank server.

Setting up these environments requires attention to prerequisites. For Slimv, the setup involves installing necessary tools like SBCL, tmux, and git, and ensuring Vim has support for the Python interface. Installation typically involves cloning the repository and placing the plugin files in a specific directory within the user's Vim configuration. Vlime requires Quicklisp to be installed first, and its setup involves cloning the repository and creating specific directory structures to ensure compatibility with Vim's package management system, sometimes necessitating manual path adjustments to the runtimepath option.

A key component shared by both plugins is Paredit, a plugin that facilitates structured editing of Lisp s-expressions by automatically managing balanced parentheses, brackets, and quotes. Slimv bundles Paredit, while Vlime requires separate installation, which is recommended for enhanced editing experience.

In terms of interactive functionality, Slimv facilitates an interactive REPL buffer, allowing direct input of code for immediate evaluation. In contrast, Vlime features a non-interactive REPL buffer by default, which poses a challenge for direct expression input; Vlime mitigates this with an interaction mode that allows evaluation upon pressing enter. Furthermore, the handling of editing features differs: Paredit’s electric return functionality works seamlessly with Slimv, mapping the enter key to insert electric returns, whereas Vlime’s mapping focuses on displaying argument lists without inserting the actual returns.

The plugins also offer distinct ways to handle advanced Lisp features. For macro expansion, Slimv uses commands like ,m for recursive expansion, displaying results in the REPL buffer, while Vlime uses commands like \ma, showing the expanded result in a separate split window. Cross-reference functionality also diverges: Slimv presents caller lists directly in the REPL, whereas Vlime presents references in a dedicated cross-reference buffer, allowing navigation directly to the referenced location.

Both plugins offer debugging and inspection capabilities. Slimv utilizes the Slime Debugger (SLDB) to show runtime errors, backtraces, and variable inspection details within the REPL context. Vlime integrates debugging through the Slime Debugger, providing a trace dialog and inspector windows that allow for inspecting variables and drilling down into the execution path, including the ability to select abort restarts.

The plugins provide various "nifty features," including evaluating top-level forms, rainbow parentheses, omni-completion, symbol description, and macro expansion. The implementation of these features varies; for instance, Rainbow Parentheses is available via a separate plugin, whereas basic symbol description is integrated differently depending on the plugin. Omni-completion works similarly in both, inserting choices upon partial entry, while the display of argument lists is managed by the respective plugin, appearing either in the status line for Slimv or in separate split windows for Vlime.

Regarding implementation, Slimv works well with Common Lisp, MIT/GNU Scheme, and Clojure, with instructions provided for setting up the interaction with CLISP and ECL. Slimv requires specific Linux environments for Scheme support and requires a particular method for locating Clojure libraries, involving manual source compilation to ensure compatibility. Vlime is restricted to Common Lisp, though it provides a specific mechanism to interface with CLISP and ECL by defining a custom function to execute the Swank server command.

The comparison reveals that Slimv is generally recommended for its broader support across Lisp dialects and its inherent interactive REPL environment, while Vlime offers a different approach, benefiting from less dependency on Python interfaces and offering specific tracing capabilities through its integration with Swank modules. Ultimately, the choice between the two plugins depends on the user's preference for the interactive nature of the REPL, the desired integration with specific Lisp implementations, and the preferred method for handling code editing and debugging.