==============================================
CSS-Challenge
Introduction
This is the Cloud Kitchens submission challenge. It was completed using Kotlin, and will run fine on any recent JVM.
Getting Started
From Code
To get started from code, open the Main.kt file and run the main function.
From Maven
To run this program using Maven, run the following commands at the project root:
mvn clean install
mvn exec:java
Technologies Used
- Kotlin
- Maven
- JDK
- Alchemy Libraries
Challenges
Top Challenges
-
The calculation of an order's value over time proved really tricky, because it depends on how long an order sits in its designated shelf vs the overflow shelf.
-
Synchronizing the shelves across multiple threads. Between the Kitchen adding orders, Drivers coming to pick up orders, and the kitchen reorganizing the shelves, there were many opportunities for race conditions.
-
Overall this was a really challenging problem. There are many complexities to this system that one would just not expect.
Handling Overflow Items
Handling overflow items is handled by the ShelfSet object,
which is an abstraction that I wrote over the four main shelves.
The kitchen adds orders to it, and the drivers pick up orders from it.
Doing it this way allows me to handle item overflow in an encapsulated manner, without troubling other parts of the system.
The ShelfSet moves orders to the overflow shelf if the current shelf is full
and there is nowhere else to go. And whenever an order is picked-up by a driver,
it checks to see if there are any items in the overflow that can take its place.
This made the ShelfSet the only point of contention across multiple threads.
In order for this to work, the ShelfSet had to be made Thread-Safe.
Areas for improvement
The result code is far from perfect. Here are some areas I have identified for improvement.
-
Dependency Injection: Dependencies between different parts of the system could have been better handled by the use of a DI framework such as Guice or Spring.
-
Real-World Dispatcher: I followed the guide, and so the Dispatcher dispatches a driver per request. In the real world it won't be this simple. Sometimes you only have a small pool of drivers doing deliveries. In fact there are a lot of complexities on the delivery side that were just glossed over for the purpose of this examination. One can imagine for instance, not just traffic, but the distance the driver must go to deliver an order. There is also a decay that happens when an order sits with a delivery person.
-
Display: For the purposes of time, I used a primary text-based display, but with more time we could have a nice UI to show the system status.
-
REST API: Right now this is just a console app. It would be nice if this thing had a REST API that allowed you to place orders and get the current system status.
-
Unit Test Coverage: Right now coverage stands at 79%, which isn't bad, but could be better.
Software Used
- IntelliJ Ultimate 😎
- Oracle JDK
- Atom Text editor
- Google Music 🎧
Hardware Used
- iMac Mid-2011 🖥
- Apple Chiclet keyboard
- Apple Magic Mouse
- Some vegan ice-cream 🍨