This docker image provides an easy way do run LilyPond inside docker.
The tags of this image correspond to LilyPond versions. So in order to use the 2.22.0 version you can use codello/lilypond:2.22.0. Not all LilyPond versions are available.
At every point there are two actively maintained versions of LilyPond. One stable and one development version. This repository follows the release cycle of LilyPond meaning there are will be two tags built based on this repository.
stableandlatestboth refer to the latest stable LilyPond version (that is the latest2.xrelease wherexis an even number)devrefers to the latest development version of LilyPond.
All images contain LilyPond, its runtime dependencies, the Open LilyPond Fonts and some commonly used build tools such as make.
Before LilyPond 2.22 was released this repository contained additional tags for each version, namely a -shell and a -basic tag. These additional tags were removed as they served little to no benefit. Now all tags contain the additional fonts as well as some common CLI tools (such as make). The default entrypoint is the lilypond binary.
To run lilypond you just need to run the image as an executable:
docker run -v $(pwd):/work codello/lilypond score.lyWhen running lilypond this way you need to pay attention to the following details:
- Map your working directory (or your sources directory) to the
/workdirectory inside the container. - Run lilypond normally using the image name instead of the
lilypondexecutable. The default entrypoint of this docker image just invokes thelilypondexecutable inside of the/lydirectory. - Any arguments following will be passed directly to
lilypond. - The output will be put into the
/workdirectory as well by default. If you use other directories remember to add bind mounts for those as well.
This docker image compiles Guile and LilyPond from source. The compilation process is quite demanding on the CPU and might take a while. Compilation via GitHub Actions takes ~8 minutes.
LilyPond inside the image is compiled into the /opt folder. This way we can take advantage of docker’s multi-stage builds and get a smaller final image. Compilation into /usr/local would work as well but would require more effort in the later stages (because we need to isolate the LilyPond files from other programs in the prefix).