calcentral: A Ruby repository from yuhunglin

#CalCentral

Home of CalCentral.

Master:
QA:

Dependencies

Installation

Install postgres

brew update
brew install postgresql
initdb /usr/local/var/postgres

For Mountain Lion users ONLY: There's a few extra steps.

Start postgres, add the user and create the necessary databases

pg_ctl -D /usr/local/var/postgres -l /usr/local/var/postgres/server.log start
psql postgres
create database calcentral_development;
create user calcentral_development with password 'secret';
grant all privileges on database calcentral_development to calcentral_development;
create database calcentral;
create user calcentral with password 'secret';
grant all privileges on database calcentral to calcentral;
create database calcentral_test;
create user calcentral_test with password 'secret';
grant all privileges on database calcentral_test to calcentral_test;

(If your PostgreSQL server is managed externally, you'll probably need to create a schema that matches the database username. See CLC-893 for details.)

Fork this repository, then:

git clone git@github.com:[your_github_acct]/calcentral.git

Go inside the calcentral repository

cd calcentral
# Answer "yes" if it asks you to trust a new .rvmrc file.

Install jruby

rvm get head
rvm install jruby-1.7.3
cd ..
cd calcentral
# Answer "yes" again if it asks you to trust a new .rvmrc file.

(Optional for development) Make JRuby faster & enable C extensions by running this or put in your .bashrc:

export JRUBY_OPTS="-Xcext.enabled=true -J-d32 -J-client -X-C"

WARNING: Do not switch between 32-bit and 64-bit JRuby after your gemset has been initialized (your bundle library will have serious issues). If you do need to change settings, make sure to reinitialize your gemset:
- rvm gemset delete calcentral
- (set your JRUBY_OPTS)
- bundle install

Download the appropriate gems with Bundler

bundle install

Copy and update the settings

mkdir ~/.calcentral_config
cp config/settings.yml ~/.calcentral_config/settings.local.yml
cp config/settings/testext.yml ~/.calcentral_config/testext.local.yml
cp config/settings/development.yml ~/.calcentral_config/development.local.yml
cp config/settings/production.yml ~/.calcentral_config/production.local.yml

and update the settings in the .local.yml files. Settings live outside of the project dir to prevent accidental commits to the repo. You can also create Ruby configuration files like "settings.local.rb" and "development.local.rb" to amend the standard config/environments/*.rb files.

Install JDBC driver (for Oracle connection) You may already have an Oracle driver from MyBerkeley-OAE development, in which case you just need to copy it to your project ./lib directory: cp ~/.m2/repository/com/oracle/ojdbc6/11.2.0.3/ojdbc6-11.2.0.3.jar ./lib/

Otherwise, download ojdbc6.jar
Copy ojdbc6.jar to your project's lib folder```

Initialize PostgreSQL database tables

rake db:schema:load

Deploy into TorqueBox

bundle exec torquebox deploy .

Start the server

bundle exec torquebox run -p=3000

Access your development server at localhost:3000. Do not use 127.0.0.1:3000, as you will not be able to grant access to bApps.

Front-end Testing

Front-end jasmine tests live in spec/javascripts/calcentral/*.

To run the tests headless on firefox run rake jasmine:ci.

To view results of front-end tests, run rake jasmine in a separate terminal, then visit localhost:8888.

Role-Aware Testing

Some features of CalCentral are accessible only to users with particular roles, such as "student." These features may be invisible when logged in as yourself. In particular:

My Academics will only appear in the navigation if logged in as a student. However, the "Oski Bear" test student does not fake data loaded on dev and QA. To test My Academics, log in as user test-212385 or test-212381 (ask a developer for the passwords to these if you need them). Once logged in as a test student, append "/academics" to the URL to access My Academics (this will change when CLC-1755 is resolved).

Debugging

Emulating production mode locally

Make sure you have a separate production database. In psql:

create database calcentral_production;
grant all privileges on database calcentral_production to calcentral_development;

In calcentral_config/production.local.yml, you'll need the following entries:

secret_token: "Some random 30-char string"
postgres: [credentials for your separate production db (copy/modify from development.local.yml)]
campusdb: [copy from main config/settings.yml, modify if needed]
google_proxy: and canvas_proxy: [copy from development.local.yml]
  application:
    serve_static_assets: true

Populate the production db by invoking your production settings:

rake db:schema:load RAILS_ENV="production"

Precompile the assets: (more info)

bundle exec rake assets:precompile

Start the server in production mode

rails s -e production

If you're not able to connect to Google or Canvas, export the data in the oauth2 from your development db and import them into the same table in your production db.
After testing, remove the static assets and generated pages

bundle exec rake assets:clean

Test connection

Make sure you are on the Berkeley network or connected through preconfigured VPN for the Oracle connection. If you use VPN, use group #1 (1-Campus_VPN)

"Act As" another user

To help another user debug an issue, you can "become" them on CalCentral. To assume the identity of another user, you must:

Currently be logged in as a designated superuser
Be accessing a machine/server which the other user has previously logged into (e.g. from localhost, you can't act as a random student, since that student has probably never logged in at your terminal)
Have enabled act_as in settings.yml (features:)

Access the URL:

https://[hostname]/act_as?uid=123456

where 123456 is the UID of the user to emulate.

n.b.: The Act As feature will only reveal data from data sources we control, e.g. Canvas. Google data will be completely suppressed, EXCEPT for test users. The following user uids have been configured as test users.

11002820 - "Tammi Chang"
61889 - "Oski Bear"
All IDs listed on the "Universal Calnet Test IDs" page

To become yourself again, access

https://[hostname]/stop_act_as

Logging

Logging behavior and destination can be controlled from the command line or shell scripts via env variables:

LOGGER_STDOUT=false - Only log to the default files
LOGGER_STDOUT=true - Log to standard output as well as the default files
LOGGER_STDOUT=only - Only log to standard output
LOGGER_LEVEL=DEBUG - Set logging level; acceptable values are 'FATAL', 'ERROR', 'WARN', 'INFO', and 'DEBUG'

Tips

On Mac OS X, to get RubyMine to pick up the necessary environment variables, open a new shell, set the environment variables, and:

/Applications/RubyMine.app/Contents/MacOS/rubymine &

If you want to explore the Oracle database on Mac OS X, use SQL Developer
We support source maps for SASS in development mode. There is a great blog post explaining how to set it up and use it.

Best Practices

In some places, we echo unescaped script tags back to the browser, so we need to be very careful not to expose those in any places where they could get executed.

Never use ngBindHtmlUnsafe.
Never use innerHTML unless displaying completely static data.

Styleguide

Use an editor that supports .editorconfig. Feel free to have a look at the editor plug-ins
Use 2 spaces for indentation
List items/properties alphabetically
Remove console.log() messages when committing your code.
Only use anchor tags <a> for actual links, otherwise use <button> instead. This is especially important for IE9.
Use single quotes when possible

👎

var name="Christian Vuerings";

👍

var name='Christian Vuerings';

Use data-ng- instead of ng- or ng: and add data- for directives

👎

<ng:view>
<span ng-bind="name"></span>
<input mmddyyvalidator />

👍

<div data-ng-view></div>
<span data-ng-bind="name"></span>
<input data-mmddyyvalidator />

Recording fake data feeds and timeshifting them

Make sure your testext.local.yml file has real connections to real external services that are fakeable (Canvas, Google, etc). Now do:

rake vcr:record
rake vcr:prettify

vcr:record can also take a SPEC=".../my_favorite_spec.rb" to help limit the recordings.
vcr:prettify can also take a REGEX_FILTER="my_raw_recording.json" to target a specific raw file.

You can now find the prettified files in fixtures/pretty_vcr_recordings. You can edit these files to put in tokens that will be substituted on server startup. See config/initializers/timeshift.rb for the dictionary of substitutions. Edit the debug_json property of each response, and timeshift.rb will automatically convert debug_json to the format actually used by VCR.

Rake tasks:

To view other rake task for the project: rake -T

rake spec:xml - Runs rake spec, but pipes the output to xml using the rspec_junit_formatter gem, for JUnit compatible test result reports
rake vcr:record - Refresh vcr recordings and reformats the fixtures with formatted JSON output. Will also parse the reponse body's string into json output for legibility.
rake vcr:list - List the available recordings captured in the fixtures.

Memcached tasks:

A few rake tasks to help monitor statistics and more:

rake memcached:clear_stats - Reset memcached stats from all cluster nodes
rake memcached:empty - Invalidate all memcached keys from all cluster nodes
rake memcached:get_stats - Fetch memcached stats from all cluster nodes
Generally, if you rake memcached:empty ( WARNING: do not run on the production cluster unless you know what you're doing), you should follow with an rake memcached:clear_stats.
All three task take the optional param of "hosts." So, if say you weren't running these tasks on the cluster layers themselves, or only wanted to tinker with a certain subset of clusters: rake memcached:get_stats hosts="localhost:11212,localhost:11213,localhost:11214"

Using the feature toggle:

To selectively enable/disable a feature, add a property to the "features" section of settings.yml, e.g.:

features:
  wizbang: false
  neato: true

After server restart, these properties will appear in each users' status feed. You can now use ng:show in Angular to wrap the feature, e.g.:

<div data-ng-show="user.profile.features.neato">
  Some neato feature...
</div>

or, depending on the feature, it may make more sense to disable it in erb (so that Angular controllers are never invoked at all):

<% if Settings.features.neato %>
  <%= render 'templates/widgets/notifications' %>
<% end %>

yuhunglin/calcentral