Spectrum

Forked from Columbia Libraries Unified Search & Discovery

Using docker-compose for development

Clone the repository

git clone git@github.com:mlibrary/spectrum.git spectrum
cd spectrum

Copy .env-example to .env

cp .env-example .env

Get the actual values for the .env file from one of the developers. Update .env with those values.

Build it:

docker-compose up --build --no-start

Install the gems into the gem-cache:

docker-compose run --rm web bundle install

Pull the latest version of the search front end:

docker-compose run --rm web bundle exec rake 'search[latest,local]'

On a fresh install it is necessary to first manually pull the catalog-solr image, which will be used to hold the local example data:

docker-compose pull catalog-solr

Load up the catalog with some example data. To do that you need to start up catalog-solr and then index the data.

docker-compose start catalog-solr
docker-compose exec catalog-solr bash /examples/load_into_solr.sh

Then start it

docker-compose up

In the browser go to http://localhost:3000

Running Tests

docker-compose run --rm web bundle exec rspec

Debugging

Puma doesn't always play nicely with pry. Worker timeouts can end sessions early, and multi-threading can make taking input from the terminal troublesome.

To address these issues, the script/docker-startup file will run in WEBrick when started with the environment variable RAILS_SERVER set to webrick. To set RAILS_SERVER add it to your .env file:

RAILS_SERVER=webrick

Additionally, WEBrick needs to run in development mode. Add RAILS_ENV to your .env to ensure it does:

RAILS_ENV=development

Changing environment variables in docker-compose gets updated when running docker-compose up so, to pick up environment variable changes every time, and attach for use with a debugger:

docker-compose up --no-start web && \
  docker-compose start web && \
  docker attach "$(docker-compose ps -q web)"

Overview of Spectrum

Spectrum is the server-side support for Search. It is intended to mediate requests to solr and summon as a back-end interface to support library Search.

Dependencies

• spectrum-config

  • Spectrum-config handles individualized configuration of the targets searchable in the front-end. It works a lot like a jello-mold. Pour the user's request into it, get a tasty treat out.

• spectrum-json

  • Spectrum-json handles receiving requests from end-users, applying any global normalization to the request, and passing it to the appropriate configuration to make a search against solr or summon.
  • Spectrum-json also handles things like taking actions on records, and identifying patron affiliation.