npm install- if you have not already done sonpm start- navigate browser to
http://localhost:35472
window.autocompleteis globally available when you add the library to a page. Using thewindow.autocomplete.init( {ContentEditableNode} )function will initiate the autocomplete bindings.- Users simply need to type
@then the user they are trying to match.
- We are using a contenteditable element as the base for our text editing. What it allowed that textarea or input[type=text] did not, was the ability to markup elements inside of itself. This allows us to style the elements and identify the user tags when we process the content on the backend.
- Decided to use usernames as the primary selection method. It is assumed to be more unique than actual names and would keep the consistency of how users reference each other. Meaning, regardless of how you searched for your suggestion, you will see the username filled into the content when selected.
- Implementation of how we should render/store this data was not in scope, but I would push that we track the user in the comment by id. On render we pull out the up-to-date value for the username. Allowing links to not become stale. Note, I updated the data to included id in the sample data.
- The search suggestion sorting algorithm is not implemented with the scope of this project. I use a very primitive check to drill down suggestions but I believe proper suggestions (Levenshtein distance calculation, mispellings, approximate matching, etc.) would be behind a backend service.
- Responsiveness and mobile readiness did not make it into the work of this phase 1.
- Frameworks. This is entirely framework agnostic and can be adapted to any if needed.
- Build process. I could have used Babel, Gulp, LESS, etc. but I felt the scale of what needed to be done did not warrant that need and would make it a little more complex than it might need for this.
- The testing has not been set up for this project. Considering how low level some pieces of the code touch, I would push for some sort of webdriver/selenium based integration testing to get the interactions properly vetted. Take those tests and put them on a continuous testing pipeline with SauceLabs to do XBrower regression testing.