Pivotal Tracker
Heroku Deployment
The goal of this currently bare-bones app, thrown together by Armando Fox with contributions by Andrew Halle, is to enable continuous tracking over time of customer apps developed by the "ESaaS ecosystem" around UC Berkeley CS169 Software Engineering.
The data initially used to populate the app came from this Google spreadsheet.
Since we have had many repeat customers who come back in subsequent semesters to have a new student cohort enhance an existing app, this system will track an app's current status over its lifetime as it is handed off from cohort to cohort.
The main models are:
- App: a deployable Web app, i.e. a student project. App statuses can fall into two categories:
- Deployment statuses:
dead
: Not deployed, and/or customer not actively using; dormantdevelopment
: In active development (a team is working on it right now), whether or not deployed in productionIn use
: In production use at a customer site; customer has not expressed interest in further improvementsIn use and wants improvements
: In production, and customer is interested in further developmentInactive but wants improvement
: An app whose current state isn't functional enough for customer to use yet, but customer is interested in further development to make app usefulPending
(deprecated): a customer has suggested an app they want built or improved, but a coach/instructor hasn't yet vetted whether it's a good fit for some student team. This has been replaced by the vetting statuses mentioned below. DO NOT use it anymore
- Vetting statuses:
Vetting
: Pending (not yet vetted)On Hold
: We need something from customer during vetting phaseStaff Approved
: Approved by the faculty during vetting phaseCustomer Informed
: Staff has approved the project and we’ve informed the customer about acceptance and are waiting for them to confirm whether they meet our customer expectationsCustomer Confirmation Received
: Customer has confirmed to meet our expectationsDeclined by Staff
: Project declined by staff during vettingDeclined by Customer
: Engagement declined by the customer after we accepted the projectDeclined by Customer – Available Next Semester
: Customer is not available this semester but will be available for next semesterBackup
: We are saving this project as a backup in case a client drops
- Deployment statuses:
- Org: a customer organization for whom the app was developed
- User: various subcategories, including developer (e.g. student), coach (mentor, GSI), customer contact. Also a principal for authentication: as of now, only a staff member has authorization to edit/destroy.
- Coaching org: an "organization" whose main function is to provide mentoring/coaching to students building apps. E.g., "UCB CS169 Fall 17" is an org, as is "AgileVentures", and so on. I would propose that each offering of CS169 be its own org, so we can track engagements accurately.
An engagement
is a period of time over which a coach interacts with
developer(s) to work on an app. During that time, the app is in
development
status. After the engagement ends, the app is either in In use
status
(customer is using it; app may be enhanced in future) or dead
(customer not using it, because it doesn't meet their needs enough to be
usable).
At any given time every app is always part of an engagement, so engagements have a start date but no end date; an engagement ends when the app transitions into another engagement.
So for example, an app that is developed in Spring 2017, used by the customer over the summer, and picked up by another cohort for enhancements in Fall 2017, might have these engagements:
Status | Start date | Coaching org | Coach |
---|---|---|---|
development | 1/15/2017 | CS169 S'17 | Tony Lee |
maintenance | 5/5/2017 | AgileVentures | Sam Joseph |
development | 8/23/2017 | CS169 F'17 | An Ju |
The goal is to have a robust ecosystem that eventually encompasses not only UCB CS169 but its offshoots: the AgileVentures volunteer-developer corps, ESaaS-like courses at other schools (Texas A&M now emulates UCB's approach and builds software for local nonprofits), etc.
When a new course offering starts, or when a non-course org is looking to source projects, they can look here to find apps in need of enhancement; if greenfield apps are built, they can be registered here so that future dev teams can pick up and enhance them.
You will need Ruby 2.3.1 and Rails 4.2.7 installed.
Clone the repo, and run rake db:setup
and then rake db:seed
to
seed the development database with a subset of the initial data set.
You should then be able to use rails server
to start the app, and
point your browser at http://localhost:3000
to access it.
In production, you login with your GitHub account. Login is only
permitted for a user whose github_uid
field in the database is set to
their GitHub username, e.g. armandofox
. So, get someone who already
has this field set to set the field for your user record.
You also need to be a coach
to navigate through the app and do some core operations
(create, update, delete). In order to give permission at database level run rails
console on heroku server(heroku run rails console
) and create/update a user:
User.create(name: 'USERNAME', email: '[email protected]', github_uid: 'username', user_type: 'coach')
The file db/github_mock_login.yml
contains the attributes for a fake
user that you can login-as for development work. You will always be
logged in as the user whose info appears in this file. Important:
You must have run rake db:seed
to create the fake orgs, apps, and this
user.
The file config/application.yml.asc
is an encrypted version of the
file containing the GitHub application key and secret for OmniAuth.
You shouldn't need to change it, but if you do, get the encryption key
from @armandofox so that you can decrypt, modify, then re-encrypt and
commit application.yml.asc
.
If you want to have GitHub OAuth on the development environment or on the heroku
deployment environment, you have to register your app here. After you register and obtain Client ID and Client Secret, add
the keys to config/application.yml
. Make sure you set the authorization callback URL to <homepage-url>/auth/github/callback
We used figaro
gem to upload app environment variables. You can add secret keys
in config/application.yml
. Important: since you are storing security-sensitive
information, remember to add this file to .gitignore
. The following keys are
needed to correctly run the app:
secret_key_base
: this is used to encrypt and sign session in order to safely send session back and forth in cookiesgithub_key
,github_secret
: these are used for login with GitHub
Although the app mocks the GitHub OAuth mechanism for test and development environment,
you still need to add a "mock key" to config/applicaiton.yml
. For example:
test:
secret_key_base: test
github_key: test
github_secret: test
development:
secret_key_base: development
github_key: development
github_secret: development
However, we think it is a good practice to have a mock key that resembles a real
key. You can easily generate a key using rake secret
.
To upload the keys to a Heroku app, run figaro heroku:set -e production
.
After setting environment variables using figaro
, you can access them by
ENV["YOURKEY"]
or Figaro.env.YOURKEY
. Refer the documentation for more information.
To enable email delivery functionality, technically you can use any third-party email delivery service as you wish. But the easiest way in this app is to register and use a Sendgrid API key by setting
SENDGRID_API_KEY: <your_sendgrid_api_key>
in config/application.yml
For local testing purpose, this is also acceptable:
export SENDGRID_API_KEY='<your_sendgrid_api_key>'
in your favorite shell.
Since Heroku wipes out all data when dyno server is down, we used AWS S3 Bucket
to store the images. After you open an account for AWS, you will need the following
keys (in config/application.yml
):
AWS_ACCESS_KEY_ID: <your_aws_access_key_id>
AWS_SECRET_ACCESS_KEY: <your_aws_secret_access_key>
S3_BUCKET_NAME: <your_s3_bucket_name>
AWS_REGION: <your_aws_region>
S3_HOST_NAME: <your_s3_host_name>
We used Cucumber/Capybara for integration tests, and RSpec for unit tests. You can run tests using:
bundle exec cucumber
bundle exec rspec
To test javascript behaviors, Cucumber uses Selenium Webdriver as default. This requires you to have a geckodriver, and Firefox browser. If you want to use other drivers (e.g. chromedriver) refer to Capybara webpage to configure default webdriver.
If you do not want to download a new webdriver, you can skip scenarios which require webdriver by:
bundle exec cucumber --tags ~@javascript
- 9 vetting statuses added to support vetting phase.
pending
should be obsolete- Staff can add vetting comments
- Apps listing page can show only the apps in vetting
- Repo URL is optional when creating a new app for vetting
- Each
app
hasfeatures
for vetting purpose. - Each
engagement
hasfeatures
for deployment purpose - Display app counts per each status on apps listing page
- The ability to "email to organizations" from single point of contact
- Support for rich text editing and display, such as bold, italic and ordered/unordered listing
- Pagination for
apps
,Orgs
andUsers
- Enhanced search bar with checkboxes. Description of each
app
is searchable - Miscellaneous:
- Org names and user names are clickable.
- Clicking
Back
button when editingapp
redirects to the app's show page instead of the apps listing page - Move
Edit App
andBack
buttons to the top right of the page - Show
app
status in the app's show page
- New
App
,Org
, andUser
can be created all at once, with proper association - Every user can "post" comments on an
App
,Org
, andUser
App
has different types of comments- Any class that inherits
Commentable
can have many comments
- More comprehensive customer feedback through a feedback form with ratings/comments
- Aggregates customer feedbacks from all iterations of an engagement, and display averages on each category
User
supports different types (e.g. Student, Staff/Coach, Customer)- Exports
Engagement
information as a CSV file - each
User
contains a profile image- we are using Amazon S3 to store images on production environment, because Heroku has ephemeral filesystem. If you want to run this app on heroku server, you will have to create another Amazon S3 account and setup the configuration(Instruction).
- Authorization to edit/destroy only to "Coach"
- Autocomplete dropdown list (select2)
- Major Bootstrap styling
- Add user contact info and a way to track user meeting notes
- Google or Facebook or LinkedIn login for customer contacts
- Manage customer feedback as a active record, not a json string
- Add multiple user types (e.g. CS169 staff can be both a coach and a client)
- Mailing customer feedback forms to customers for each iteration (Sendgrid)
- More authorizations to different types of users
- A user cannot edit/delete other users unless it is a staff/coach