edcontext

This project contains three pieces:

• A webapp for visualizing and understanding MCIEA survey data
• Tasks for processing and indexing raw survey data
• Experimental work to conduct survey samples via text message

This is a Rails project, deployed on Heroku.

Site links

Live app: https://mciea-dashboard.herokuapp.com/ Live dashboard: http://mciea-dashboard.herokuapp.com/districts/winchester/schools/winchester-high-school/dashboard?year=2020-21

Local development

Install Postgres and running first.

(MacOS, Optional), you can use Homebrew:

brew install postgres
brew services start postgresql

Linux:

Install postgres. Known working version is version 13

https://www.postgresql.org/download/

On linux, if you run into problems creating the postgres database, edit /etc/postgresql/13/main/pg_hba.conf. Change the connection method of IPv4 and IPv6 connections from md5 to trust.

# IPv4 local connections:
host    all             all             127.0.0.1/32            trust
# IPv6 local connections:
host    all             all             ::1/128                 trust

Once postgres is installed and running, install the required gems and then migrate the database.

bundle install
bundle exec rake db:create db:schema:load db:seed

Install the javascript dependencies

yarn install

At this point you can run the app and login. There won't be any data yet though; keep reading!

The seed file populates the following tables

Name	Description
School	School ids are only unique to their district. More than one school has an id of 1
District	Districts and schools have attached slugs. We find search for these models by their slugs
SqmCategory	The legacy name here is Category. It still exits in the database. We wanted the freedom to make changes and still preserve the legacy site until the end of the engagement.
Measure	In the bar graph measures represent a single bar
SurveyItem	This table has an attribute prompt that is the question asked

SurveyItemResponses does not get populated at this stage.

Database

Postgres

Gems

Name	Description
puma	webserver
pg	postgres
jquery-rails	legacy, allows use of jquery
jbuilder	legacy, build json objects
haml	legacy, write views in haml syntax
bootstrap	css framework
newrelic_rpm	legacy?, application monitoring
devise	authentication
omniauth	authentication
twilio-ruby	legacy, text messaging
activerecord-import	faster database imports
stimulus	Create javascript controllers
turbo-rails	Manages what gets rendered on the frontend and when
redis	Caching system
jsbundling-rails	Bundle javascript asssets
cssbundling-rails	Bundle css assets

External APIs

None yet. Hoping to integrate with Powerschool and Aspen for school administrative data.

Javascript

Esbuild is used as the javascript bundler. Scripts for esbuild are defined in package.json e.g. yarn build. This script will run if in development with bin/dev. The javascript testing library is jest. Manually run test with yarn test. Javascript tests will also run with bundle exec rake.

Stimulus is installed. Create a stimulus controller with ./bin/rails generate stimulus [controller]. If you create a stimulus controller manually, you can add it to index.js with the command stimulus:manifest:update.

css

Bootstrap 5

Loading Data

Loading Survey Item Responses

SurveyItemResponses is the most important table to understand. SurveyItemResponses is the data that will change year to year and makes up the majority of the database records. Roughly 500,000 SurveyItemResponses per year.

Some notes:

• The data loading task assumes that the CSV files live in the #{RAILS_ROOT}/data/survey_responses directory
• The data loading task is idempotent, i.e. it can be run multiple times without duplicating previously-ingested data

How to run the data loading task:

# locally
$ bundle exec rake data:load_survey_responses

# on heroku staging environment
$ heroku run:detached -a mciea-beta bundle exec rake data:load_survey_responses

# on heroku production environment
$ heroku run:detached -a mciea-dashboard bundle exec rake data:load_survey_responses

For convenience, you can use the following script for loading data on Heroku:

# on heroku staging environment
$ ./scripts/load_survey_responses_on_heroku beta

# on heroku production environment
$ ./scripts/load_survey_responses_on_heroku dashboard

There is also an example one-off task to load a single csv at a time.

bundle exec rake one_off:load_2018_19_student_responses

Loading Admin Data Values

Loading admin data is similar to loading survey item responses. Run the one of the following scripts to load admin data to a selected environment.

# locally
$ bundle exec rake data:load_admin_data

# on heroku staging environment
$ heroku run:detached -a mciea-beta bundle exec rake data:load_admin_data

# on heroku production environment
$ heroku run:detached -a mciea-dashboard bundle exec rake data:load_admin_data

Load Response Rates

Many parts of the site rely on the response rate table being populated. The response rate is taken into account when determining if there is sufficient data to show

# locally
$ bundle exec rake data:reset_response_rates

# on heroku staging environment
$ heroku run:detached -a mciea-beta bundle exec rake data:reset_response_rates

# on heroku production environment
$ heroku run:detached -a mciea-dashboard bundle exec rake data:reset_response_rates

Load Student demographic information

Student demographic information is needed for the Analyze page disagregation

# locally
$ bundle exec rake data:load_students

# on heroku staging environment
$ heroku run:detached -a mciea-beta bundle exec rake data:load_students

# on heroku production environment
$ heroku run:detached -a mciea-dashboard bundle exec rake data:load_students

Load scores for each race

For performance, the scores for an individual race must be precalculated. Make sure to load response rates, and student demographic information before recaculating scores by race

# locally
$ bundle exec rake data:reset_race_scores

# on heroku staging environment
$ heroku run:detached -a mciea-beta bundle exec rake data:reset_race_scores

# on heroku production environment
$ heroku run:detached -a mciea-dashboard bundle exec rake data:reset_race_scores

Running tests

Single threaded test execution

Prepare the test database.

bundle exec rake db:test:prepare

If you need to look at the rails console for the test environment

RAILS_ENV=test rails c

Run the tests

bundle exec rake

Parallel test execution

The parallel tests gem is installed. It's optional to use.

Set the TEST_ENV_NUMBER environment variable. For example, add this line to your .bashrc

export TEST_ENV_NUMBER="20"

Create the additional databases

bundle exec rake parallel:create

Run the tests in parallel

bundle exec rake parallel:spec

Run the tests with a specific number of processes

bundle exec rake parallel:spec[5]

Viewing test coverage

xdg-open coverage/index.html

Javascript tests

Run the javascript tests

yarn test

Connecting to Heroku git

To add the heroku remote repository for beta run git remote add beta https://git.heroku.com/mciea-beta.git

To add the heroku remote repository for production run git remote add dashboard https://git.heroku.com/mciea-dashboard.git

Continuous Integration

Pushing commits to the main branch triggers auto-deployment to the staging environment. Use the ship-it script from the main branch when you're ready to deploy to staging

scripts/ship-it.sh

Deployments to production must be done through the Heroku web interface or via the Heroku command line

Running the development server

Start esbuild for dynamic compilation of javascript assets.

yarn build --watch

Start the puma web server

bin/rails s