⚠️ Quick and dirty code here: This project is my submission to a coding challenge.It was written on a strict timelineIt was written with some haste, but mostly I wanted to have fun. I'd be more than happy to discuss any part of this solution or my thoughts on the take-home challenge as presented to me.
WhySpace is a barebones social media clone. It's a Ruby on rails project which uses Devise for login and Bitly as a link shortener to create member profiles.
You will need a Bitly API key before getting started.
- Install Ruby 2.7
- Install Bundler 2.1.4
bundlebin/rake db:setupexport BITLY_API_KEY=...bin/rails s
There are a number of seed users which can all be accessed with
login: email_#{n}@example.com
password: test1234
I wrote this over the course of 3 days, devoting around 4-5 hours per day. I have avoided rebasing as much as possible to keep the timestamps clear but have also pushed separate branches to show the major milestones.
- Rails new
- Devise for login
- Member model to track names
- Website model to track website urls
- Passion model to track topics of interest
- Profile pages #show and #my_profile to display this info
- Piggyback on registration form to collect name and website
On day1 my suspicions were confirmed. There was zero chance I'd be completing the challenge in the 4 hour time limit or anything close. I cut corners from the start, but I wasn't going to have any fun breaking my back for a speed run. So although the UI is barebones, there do exist some specs to help my process.
- Friendship model to track which Members add which other Members as friends
- Project seeds
- Add Friend form on my profile
- Show friends on profile pages
- *start work on expert search
On day2 I sorted out how I wanted friends to work. I also started imagining what an "expert search" was going to look like and played around with a couple graph pathfinding gems.
sinetheta/expert-search branch
- Friends graph
- Expert search
- Pathfinding on the graph to those experts
- Search form on my profile page
- Search results display as profile page links
- Members list page
Day 3 is when I tied it all together with the experts' search. I guess you could complain that the only computer science in this challenge I have delegated to someone's ancient 5 star pet project thinkphp/dijkstra.gem but I'm no CPSC phd and I'm more than happy to leave bit crunching to someone else. My thanks to thinkphp for putting it out there and wrapping it as a gem, because pickings were pretty sparse when it came to Ruby implementations of pathfinding algorithms.
There are some other algos which might have slightly better performance characteristics. Especially the class of solutions that connect more points simultaneously because our problem required repeated application of Dijkstra to get all the output I needed.
Ultimately this type of strategy would never scale. Before playing Kevin Bacon there's no way to know if any particular person is going to be needed, so this search is going to grow fast. Rather than try to optimize any particular search algorithm I'd be inclined to look to a more tailor suited tool like Neo4j.