scip-ruby

Experimental SCIP indexer for Ruby, enabling precise code navigation for Ruby — Go to definition, Find references, hover docs etc. – in Sourcegraph.

If you have any questions, bug reports, feature requests or feedback, please file an issue. The most common indexing workflows are discussed below; you can also consult the CLI reference docs for more information.

scip-ruby builds on top of the Sorbet type-checker. It is primarily meant for use in projects which have started adopting Sorbet. Higher Sorbet adoption (# typed: true or higher) is likely to lead to a better code navigation experience. However, scip-ruby can also index # typed: false files on a best-effort basis. Like Sorbet, scip-ruby treats files without a # typed: sigil as implicitly being # typed: false.

Currently, we have gems and binaries available for x86_64 Linux and arm64 macOS.

This section covers the easiest way to use scip-ruby: as a gem, which includes a platform-specific scip-ruby binary and uses that for indexing. Alternately, you can directly download a binary and index your code.

If you have a .gemspec file, use add_development_dependency:

Gem::Specification.new do |spec|
  # ... other stuff ...
  spec.add_development_dependency("scip-ruby")
end

Otherwise, add this line to your Gemfile:

gem 'scip-ruby', require: false, :group => :development

After either of those steps, run bundle install to download and install fetch scip-ruby.

Run scip-ruby along with some information about your gem.

• If you have a sorbet/config file, that will be picked up automatically to determine which files to index.

  bundle exec scip-ruby

• If you don't have a sorbet/config file, add an extra path argument to index all files in the project.

  bundle exec scip-ruby .

These commands will output a SCIP index to index.scip. Any other needed information will be inferred from directory layout. For customizing how scip-ruby interprets your configuration, see the CLI reference.

The SCIP index can be uploaded to a Sourcegraph instance using the Sourcegraph CLI's upload command.

If you're curious about the internals of the index, such as which files were indexed, check out the SCIP CLI and the SCIP development docs.

You can download a release binary and run it directly, similar to invoking Sorbet.

ARCH="$(uname -m)" \
OS="$(uname -s | tr '[:upper:]' '[:lower:]')" \
bash -c 'curl -L "https://github.com/sourcegraph/scip-ruby/releases/latest/download/scip-ruby-$ARCH-$OS" -o scip-ruby' && \
chmod +x scip-ruby

# If using in CI with 'set -e', make sure to wrap the
# scip-ruby invocation in 'set +e' followed by 'set -e'
# so that indexing failures are non-blocking.
./scip-ruby

The generated index can be uploaded to a Sourcegraph instance using the Sourcegraph CLI (v3.43.0 or newer). Typically, the command looks like:

SRC_ACCESS_TOKEN="your token" SRC_ENDPOINT="url for Sourcegraph instance" src code-intel upload -file=/path/to/index.scip

For more details, see the Sourcegraph CLI docs.

See the Contributing docs for build instructions. Once the scip-ruby binary is built, you can index it as described above.

See the Contributing docs.