This is a de-risking prototype supporting the implementation of a new case management system for the Department of Justice's Civil Rights Division (DOJ CRT).

Objective: Learn as much as we can about CRT's reporting needs and the constraints a future data model must support.

Method:

• Start with CRT's current data model and sample data.
• Use a variant of the red-green-refactor loop, but for reports and data, to iterate on the data model:
  • Write acceptance tests that ensure that CRT's current reports are returning the correct data.
  • Write acceptance tests for yet-to-be implemented reports, including reports that we know the data model cannot support.
  • Alter the data model (through Rails migration steps) and report queries to support the new reports while maintaining support for the existing reports.

We are specifically not using 18F's Rails template because there are no plans to deploy this code. This prototype is entirely for the purposes of gathering information, informing decisions, de-risking, and resolving ambiguity.

1. Clone this repository
2. Follow the Oracle installation instructions
3. Run bundle install to install dependencies
4. Follow the next section on setting up an Oracle database

This project has two databases: Oracle and Postgres. We kept running into the Free/Express Oracle size limits, so we do our data migration work by copying and transforming data from Oracle into Postgres.

If you have a database export from an Oracle database, you can import that to the Oracle database here.

1. Get a copy of the database export file (ending in .dmp) and a copy of the oracle_setup.sql script from a teammate. Make sure these files go into the project's db/data/ folder.

2. Start Docker Compose: docker compose up

3. Check if the db/data volume mounted properly. You should see the same files using this command as you do in the application's db/data folder.

$ docker compose exec oracledb ls /db/data

4. Set up the database. First log in:

$ docker compose exec oracledb sqlplus sys/password@XEPDB1 as sysdba

Then run the SQL in db/data/oracle_setup.sql by copying and pasting it into the database console.

5. Import the data

$ docker compose exec oracledb impdp system/password@XEPDB1 directory=db_data dumpfile=yourFileName.dmp

6. Migrate the Postgres database: rails db:migrate:modern. Migrations and import are a bit of a process right now, so ask for help.

When adding a new data migration, split the migration into two: a structure migration and a data transform migration.

Structure migrations will creating tables and editing columns, and data transform migrations involves testing that data has been set up in the revised structure. As it's set up now, you'll migrate, run into a failing test, then open a console and run the data transform step, then continue migrating. It's messy but it's a prototype and we'll live with it for now.

Just ask for help when getting started, it's a little word-of-mouth right now.

1. Run bin/acceptance to run acceptance tests. By default, they run for the development environment. Prepend the command with ENV=production to run the acceptance tests with production expectations.
2. Run rails test to run unit tests

1. Start a branch git co -b <simple-description-of-branch>
2. Commit to the branch. Make sure your code is reasonably well-tested.
3. When you're ready, push to the branch and create a pull request

To add a report:

1. In app/queries, create a new SQL file.
2. Save the file, giving it a filename that describes the report. If it's a section report, start it with the section abbreviation (e.g. ELS).
3. Paste in the SQL code that produces the desired report. Reports must only produce a single table of output. If you have multiple tables in a report, save specify the table in the filename after the report name, like {section code} {report name} Table 1.sql.
4. In the SQL file, replace variable literal values with Mustache tags. Use simple snake_cased variable names. These variables will be populated with values in the acceptance tests. As an example, replace the variable values in the original SQL:

var p_startyear smallint
exec :p_startyear := 2012;

var p_startquarter smallint
exec :p_startquarter := 4;

WITH charged_hours AS (

    SELECT # ...

with simple variable names like

var p_startyear smallint
exec :p_startyear := {?start_year};

var p_startquarter smallint
exec :p_startquarter := {?quarter};

WITH charged_hours AS (

    SELECT # ...

5. Make sure the report made it in. Run rails console and then

Report.all.map(&:name).grep /a snippet of the report title/

6. Check the results by running the report:

Report.find({ title or ID }). # Find a report by its full name or by its position in Report.all
  with(required_var: value).  # Give it all the values it needs for evaluation. See a report's variables with Report#variables.
  results.first(10)           # To keep displayed results to a minimum

7. Add expectations for a report by writing acceptance tests. Replace literal expectation values with tags like {{ section.report.expectation_name }} and fill in the values for development and production in features/test_values.yml.