Etappe is a browser app for planning a metro transit trip using one or more transit agencies. There are many strategies for creating an optimal trip. Some of the optimization criteria are:
Many transit agencies (agencies) publish route information and real-time schedules via web service interfaces (APIs). Although there is a standard protocol, it does not appear to provide the real time data this application requires.
The key data structure is a trip. A trip has a origin place and a destination place. It also may have an anchor time and anchor time mode. Thus, for example a trip could depart the origin at a particular time or arrive at the destination at a particular time. Currently, a trip starts now at the origin.
When a trip is evaluated, it creates plans. Each plan is a sequence of segments, such that each plan satisfies the trip, but may use different routes, stops, and vehicles. The intention is that plans be optimized according to a strategy, which may be left up to the user.
A key subordinate data structure of a segment is a ride. A ride provide concrete instances which satisfy a segment, including details origin and destination time, carrier, and vehicle id.
http://www.bart.gov/schedules/developers/api
http://www.nextbus.com/xmlFeedDocs/NextBusXMLFeed.pdf
This project is developed using BDD and TDD. Protractor is used for BDD and Karma is used for TDD.
Node and npm are required for testing and building. You will need the latest versions. Current development uses:
The remaining toolchain is specified in the package.json file. Install by running:
npm update
One more step required for protractor is to install the webdriver manager:
node ./node_modules/protractor/bin/webdriver-manager update
It is assumed you already have the latest web browsers installed. Most of the development has been with Chrome browser.
The unit tests and integration tests can be run from the command line. I tend to run the test in IntelliJ, because the workflow is easier in the IDE.
The unit tests are run with Karma.
There is a script in the package.json file to simplify this.
Just run:
npm test
The karma configuration invokes the spec reporter, so you'll see a nice summary of the tests in the command output.
Running the Protractor integration tests from the command line is a bit more complex. Three command line terminal windows are required to run: 1) the web server, 2) the Mountebank server, 3) the Protractor tests.
In the first terminal window, run the web server:
npm start
In the second terminal window, run Mountebank:
npm run mountebank
In the third terminal window, run the Protractor tests:
npm run protractor
Protractor also produces a spec summary of the tests, similar to the unit tests.
If you are using IntelliJ, here are some additional setup steps for running the tests within the IDE.
IntelliJ has plugins which make TDD with Karma and Protractor highly productive.
The IntelliJ Karma plugin makes Javascript testing really nice. First make sure you have the Karma plugin installed. If not, it's easy to do:
Preferences > Plugins > Install Jetbrains plugin. Navigate to and select the Karma plugin. Click Install and restart.
Once you have installed the plugins setup the following run configurations:
protractor tricks for running in jenkins
Launch each of the four tools and pin them in the Run window. This includes the web server and mountebank, which will be always running. Run the Karma unit test or the Protractor E2E tests by selecting the tab and clicking the run button.
Enter name, config file (it finds it), browser (start with Chrome)
Test environment
run configrations for:
details here karma, http server, mountebank, protractor
Configure Karma with the "Run/Debug Configurations" window.
Configure an Http server with the "Run/Debug Configurations" window.
Then you can go to http://localhost:8080/app/index.html
selenium driver jar issue
may need to update chromedriver: https://www.swtestacademy.com/install-chrome-driver-on-mac/
From time to time, it's necessary to update compoents of the test environment. Following are some of the common issues.
You may encounter this error when running the Protractor tests. ChromeDriver depends Chrome installed on the system. Chrome usually updates automatically, therefore it can get ahead of the ChromeDriver version. The solution is to manually upgrade ChromeDriver as follows:
sudo mv ~/Downloads/chromedriver /usr/ocal/binThe Protractor tests are not fully end-to-end because the data that would come from the transit servers varies over time.
A consistent test fixture is provided by Mountebank. Before running Protractor, start up Mountebank.
The Protractor tests inject the test fixtures into Mountebank. The Protractor beforeEach() function injects a mock module to
configure the back end service URL to the Mountebank imposter port on localhost.
Set up a node run configuration for Protractor. Initially, run it from the Run menu or toolbar. After that, a run can be easily started with the Run button in the Run pane.
In keeping with BDD, a UI story is first expressed in Protractor. That way, the BDD red-green-refactor methodology can be followed nicely.
Continuity with TDD: The author is still exploring the tradeoffs between top-down BDD and bottom-up TDD methodologies. The emphasis has been on BDD because it addresses the user story much more directly that TDD. So far, TDD comes into play when user stories demand new data models and services, or when an existing 'ad-hoc' solution demands refactoring of the code. The BDD cycle time is of course slower than TDD. The TDD tests are taking less than a second, while the BDD tests in the Chrome browser are taking about 7 seconds.
Manual testing should be done periodically to ensure the UX is reasonable. However, whenever JavaScript errors are encountered in manual testing, the error should immediately be exercised in Protractor. For example, there was an Angular injector error due to a missing script tag. By adding an Angular exception handler, the Protractor run was made red. Then the missing script tag was added to index.html to make the test green.
When manual testing exposes UX issues, they may or may not involve BDD or TDD. For example, if it's necessary to modify the DOM or add CSS classes to DOM elements, the changes should first be added to BDD tests. When other changes are made to the DOM, use protractor for regression testing. However, changes to the CSS would be manually tested changes.
Really good article about the who, why, and how of testing: https://testing.googleblog.com/2015/04/just-say-no-to-more-end-to-end-tests.html
How to mock providers: http://www.syntaxsuccess.com/viewarticle/how-to-mock-providers-in-angular
Here are some issues that may arise during development, testing, and usage.
This is likely due to not installing the webdriver manager. Refer to the toolchain setup section for instructions.
Currently the app does not validate that routes or rides exist between one chosen waypoint and the next.