A slug is a unit of deployment which includes everything an application needs to run. The slug compiler is a program which uses a buildpack to transform application source into a slug, typically by gathering all dependencies and compiling source to binaries, if applicable.

As inputs it takes a source directory, cache directory, buildpack URL, and output directory. It places a tarball of the slug as well as a JSON file of process types in the output directory if compilation is successful.

Note that this is a trimmed down version of the slug compiler that currently runs in production on Heroku. It's intended for next-gen services but is not currently in production use.

In the typical git-based Heroku deploy, the slug compiler is invoked via a Git pre-recieve hook. However, this code has been extracted so that the same process can be used elsewhere.

Note that it will delete .slugignore patterns and other files, so you shouldn't run this straight out of a checkout. Create a fresh copy of the application source before compiling it.

• Ruby 1.9
• tar
• du
• git (optional, used for fetching non-HTTP buildpacks)

The new slug compiler does much less than the old one. In particular, these operations are now the responsibility of the caller:

• checking size of the slug
• stack migration
• detecting among set of default buildpacks
• git submodule handling
• special-cased bamboo operations
• calling releases API
• storing slug file in S3
• deploy hooks