How to Hack on lem Itself

There are two ways to hack/develop lem.

• Start Swank on lem and connect from Swank client
• Directly modify lem process behavior

M-x start-lisp-repl
CL-USER> (swank:create-server :dont-close t)

M-x slime-connect

Using this method, it is safe to break lem because the editing process is separate from the main lem process.

M-x start-lisp-repl
CL-USER> (micros:create-server :dont-close t)

M-x slime-connect

M-x slime-self-connect

This will open a connection to Lem itself, the only problem with this approach is that if you are tweaking some internal features it may broke the current editor, so be careful.

If you have problems launching lem (maybe for a display error), the lem function has a nice parameter to log the information into a file:

CL-USER> (ql:quickload :lem)

CL-USER> (lem:lem "--debug" "--log-filename" "lem.log")

This should save the logs in the file “lem.log” from the directory that the the REPL was launch.

The project’s instructions show how to load and run Lem with Roswell. Here’s how we can do without, with a regular SBCL REPL and Quicklisp.

This was tested on LinuxMint 20, with SBCL 2.14 from Roswell and a recent enough Roswell Quicklisp dist (ros update quicklisp).

Clone the repository with git submodules:

git clone --recursive <url to lem>

lem uses an asd file for each of its submodules, so you can’t compile the main lem.asd and quickload it. You need to set CL_SOURCE_REGISTRY=/path/to/lem// before starting the repl. Two slashes at the end tell ASDF to search asd files in subdirectories.

$ cd lem/
$ CL_SOURCE_REGISTRY=$(pwd)// rlwrap sbcl
* (asdf:load-asd "/path/to/lem.asd")
* (ql:quickload "lem")
[…]
* (lem:lem)
[… builds and runs the ncurses frontend …]

Load the SDL2 frontend:

* (ql:quickload "lem-sdl2")

To build a binary, see the following section.

From the root directory of the Lem project, using Make.

• With ncurses:

 make ncurses

• With sld2:

 make sdl2 

This creates a binary on the root of the lem project called “lem”.