Skip to content

Latest commit

 

History

History
74 lines (53 loc) · 4.54 KB

docs.md

File metadata and controls

74 lines (53 loc) · 4.54 KB
Raw

Add any on the fly documentation in this file.

Schema usage

  • The ggircs_portal schema is the public schema. It contains all database entities that should be exposed to the api. ie search functions, tables, views etc
  • The ggircs_portal_private schema is the private schema. It contains functions that are necessary for the operation and security of the portal, but should not be exposed via the public api.

Graphile-worker

  • The workers run independent of the main app (see dev or start in package.json)
  • Graphile-worker requires a 'tasks' folder that contain javascript task files describing the actions that task should carry out
  • Calling graphile worker is done with an SQL statement like:
  • select graphile_worker.add_job('<task(filename in tasks folder)>', json_build_object(key1, value1, key2, value2...));
  • Full graphile-worker docs: https://github.com/graphile/worker

Cypress-Axe

  • Cypress-Axe is used to check accessibility issues during our end-to-end tests.
  • A cy.visit() call must come before injecting cy.injectAxe()
  • The function to make an accessibility check is cy.checkA11y()
  • Accessibility errors will show up in the cypress panel
  • Details about the errors and how to fix them will appear in the dev tools console when those errors are clicked on in the cypress browser
  • Extra config details: https://www.deque.com/axe/axe-for-web/documentation/api-documentation/#api-name-axeconfigure

Cypress: Ignore email test

  • If you want to ignore the email test locally add an env flag to the command yarn cypress --env NO_MAIL=true
  • yarn test:e2e --env NO_MAIL=true also works

Cypress: Fixtures

  • The data mutations done in cypress tests are passed from test to test
  • To help ensure each test passes or fails independent of other tests we run truncate ggircs_portal.application restart identity cascade in the setup fixture for each test. Then create 2 applications with the create_application_mutation_chain function.
  • Any specific updates to other child tables of application (ie form_result) necessary for the individual can be done once the new applications have been created.
  • This makes it more clear what data is being tested in each test and helps to avoid data changes "bleeding" into subsequent tests & causing false failures / successes.

Authentication

We use Keycloak to manage authentication of users (using username/password or IDIR) and define their privileges using groups. The Keycloak realm configurations are available in app/keycloak-*-realm-export.json.

Our application is deployed in the dev, test, and prod environments. Because we restore the prod data into the test environment, both deployments must share the same keycloak instance (so that the users uids are identical). Our dev deployment uses a separate keycloak instance to avoid polluting the production keycloak realm with fake users (although that will inevitably happen when performing QA in test). The mapping between the Openshift namespaces and the keycloak namespaces is defined in app/data/kc-namespace-map.json

Mailhog

  • MailHog can be run via docker with:
  • sudo docker run -d -p 1025:1025 -p 8025:8025 mailhog/mailhog
  • This starts a server on localhost:8025 & uses smtp port 1025

  • Running locally:

  • There are several ways to install MailHog locally detailed here: https://github.com/mailhog/MailHog
  • These are the steps I took to install it locally (Ubuntu 16.04):
  • These steps require go-lang. If you already have go set up. Skip ahead to Install MailHog

Install Go:

  • sudo apt install golang-go
  • Navigate to home directory
  • mkdir gocode
  • echo "export GOPATH=$HOME/gocode" >> ~/.profile
  • source ~/.profile

Install MailHog:

  • go get github.com/mailhog/MailHog
  • sudo cp /home/{user}/gocode/bin/MailHog /usr/local/bin/mailhog --> replace {user} with your user
  • You should now be able to run mailhog from your command line with the command: mailhog
  • MailHog can be configured in a variety of ways: https://github.com/mailhog/MailHog/blob/master/docs/CONFIG.md
  • By default command mailhog should start the server with:
    • SMTP server starts on port 1025
    • HTTP server starts on port 8025
    • Uses in-memory message storage

Metabase

Metabase needs the function ggircs_portal_private.add_enum_comparison_operators to be run for postgresql enum types to work properly in filters. There is still the downside that a search box does not contain a dropdown with the set of enum values, enum values must therefore be known in advance. Open issue: metabase/metabase#7092