The purpose of this project is to speed up initializing a new Rails project. Our agreed upon list of Rails gems and settings are included so that we don't need to discuss each choice that has been made in the past.
- Clone this project from Github
- have a repo ready for the new project
- make a new directory with the new project name, cd into that directory
git clone git@github.com:CKDev/greenfield.git .(with the .)- update the remote links from greenfield to the new project
git remote rm origingit remote add origin git@github.com:<new-project-repo>.git
- Change .ruby-gemset for new project name
- Create a new gemset with 'rvm gemset create ' or just cd out and back into the directory
gem install bundlerif not in your global gems- Run
bundle install - Update database.yml.example with your credentials, rename to database.yml
- Update secrets.yml.example with valid SendGrid credentials, rename to secrets.yml
- Create a new secrets key in secrets.yml with
rake secret - Run
rake db:create db:migrate db:seed - Fix all TODOs in code, as there are several spots that need project specific information.
- Find all locations of "greenfield" and replace.
- Configure Redis - TODO:
- Configure Sidekiq - TODO:
- Make sure the :repo_url is updated in deploy.rb.
- Make sure the staging and development servers are ready, and that you can ssh into them.
- Run cap staging deploy once to setup the directory structure. This will fail for a few reasons, but it helps to have a starting point to work with.
- Login to the staging server.
- Create the secrets.yml file with appropriate stage.
- Go into any of the releases directories (there's probably only one, and current won't be set up yet).
- Run
RAILS_ENV=staging bundle installto get all dependencies. - Run
RAILS_ENV=staging rake db:createto create the initial database.
- Run
- Back locally, run
cap staging deployagain. This should have better results. - Update the secret_key_base in secrets.yml on the server, by running
RAILS_ENV=staging rake secret - Repeat for Production.
2.3.1
4.2.7
- Redis
- Postgresql
- Update database.yml.example with your credentials
- Run
rake db:create db:migrate db:seed - Run
RAILS_ENV=test rake db:create db:migrate
Rspec is used on this project, which can be run with: rspec
- Sidekiq
[Info on deployment]
(E.g. from the Showami Project, replace with this project's norms)
Deployment is done via Capistrano
cap staging deploy
cap production deploy
Note that sidekiq is required for this app to be functional, and therefore should be verified after deployment.
On the server run ps aux | grep sidekiq and verify that the service is running.
The purpose of this section is to layout the norms of this project. Future development should follow the standard set forth in this guide.
Rubocop is used on this project, which defines the Ruby styling agreed upon for this project. The rules are bendable, but a best effort should be made to stay within the rubocop checks. At the time of MVP, the Rubocop checks all passed.
[Info on JavaScript testing/code standards and norms]
(E.g. from the Showami Project, replace with this project's norms)
At this time there is no JavaScript testing or linting, as there is simply not enough JS code in the app to justify the effort. This should be reassessed over time.
[Info on Rails testing/code standars and norms]
(E.g. from the Showami Project, replace with this project's norms)
This project was test driven from the start, and any new features or bug fixes must have an accompanying test, or a valid reason as to why a test isn't possible. At the time of the MVP the testing coverage was > 98%.
A feature test to prove the actual working feature is preferred. Edge cases aren't necessary with feature tests. From that, more granular controller and model testing to cover different code paths and edge cases is ideal.
At any time, the working state of the app should be provable by running the test suite.
[Info on deployment norms]
(E.g. from the Showami Project, replace with this project's norms)
I am following a simple branching strategy. Master at this time is the main branch, and is deployed to staging for review. Developers should use feature branches for development, but then merge to master for review. The Production server environment maps to the production github branch.
I'm following a tagged release strategy, loosely based on SemVer. Master should be tagged, using SemVer, and then the cuts of the production branch can be made a specific tag points, with the release notes being the oneline commit titles from the previous tag.
For example:
- First get release notes (in a different tab)
git log --onelinegit tag -a vx.x.x(Add title for release, then paste in release notes from above step)git push origin vx.x.x
Git commits are like any other piece of code, and should be done with intention. There are two parts to the commit - the title and the body. The title in Github is limited to 50 characters, so the first line of a commit should also be limited to 50 characters. The body is limited to 72 characters in width, make sure your lines are no longer than 72 characters.
More importantly, a title should have a tag like [CHG], [FEAT], [REFAC], [BUG] etc, so that when a release is made, the corresponding changes are all easily visible. The body of a commit should list the why, not the how. The how should be obvious by the corresponding code changes. The title should be in the active voice, i.e. "Change timeout to two hours", not "Changes timeout to two hours." An easy way to remember this is that the commit title should finish the sentence, "If I pull in this change it will ..."
Commits should be "squashed" into atomic chunks of code, usually corresponding with a full feature or change. WIP commits are not within the code standards of this project. Any checkin should be deployable, without having to consider the surrounding commits.
...