FizzBuzz

FizzBuzz REST API over HTTPS (example).

This project offers 2 endpoints:

• A simple package to play with the original FizzBuzz or make your own one. Three algorithms are available.
• A REST API over HTTPS to make your own FizzBuzz and get its responses as JSON.

Installation

Fizzbuzz uses dep to manage its dependencies. Waiting for Go Modules (code name: vgo), it's the tool to do this stuff!

$ go get github.com/rvflash/fizzbuzz
$ cd $GOPATH/src/github.com/rvflash/fizzbuzz
$ dep ensure

Quick start

If you just need a Go package to play with Fizzbuzz, see this documentation on the interface. This package proposes 3 different methods to apply the algorithm. The benchmark's performances have defined the method to use in the API.

import (
	"fmt"
	"github.com/rvflash/fizzbuzz"
)
// ...
fb := fizzbuzz.Default
fmt.Println(fb.Bulk(15))
// output: [1 2 fizz 4 buzz fizz 7 8 fizz buzz 11 fizz 13 14 fizzbuzz]

Otherwise, to start the REST API, see the cmd/fizzbuzz for more information. Also see, the documentation of the API handler.

Dependencies

Gin

One of the best features of Go is its built-in net/http library. However, I wanted to use Gin for this project to simplify some parts of the code and used a non-standard package to use dep. And, it's a really fast alternative to build a micro-service like this one (see Goa for more complex projects).

Features

• A package that exposes methods to play with FizzBuzz, with 3 different algorithms (see benchmark bellow).
• A REST API over HTTPS to use it.
• Continuous integration with Travis that run all lints and tests (benchmark included).
• Fully tested and code that passed various linters (go vet, etc.)
• A Systemd unit file to easily manage and control the service. The logs can be manipulated with Journalctl.
• The API can recover on its own panics. Systemd will restart it for others cases (server reboot, etc).
• The API exposes a health check as monitoring purpose.
• Vendored dependencies with dep.
• Semantic Versioning. The master branch, as usual with a Go project is not safe. To use in production, see the latest tag: v0.0.2.

TODO: possible improvements

• Add a cache. As the result is deterministic, we can use a cache to speed up operations and store the API results.
• Add the cache-control header to also cache the data on client-side (max-age).
• Add the CORS headers (Access-Control-Allow-Origin, etc.) to make the API accessible by JavaScript in-browser client-side code.
• Add a system, via NGINX or other, to redirect the HTTP traffic to the HTTPS API.
• Add GZIP encoding to optimize the load time.

Testing

As usual:

$ go test -cover -race -v $(go list ./... | grep -v /vendor/)
$ go test -bench=.

Benchmark

On Linux OS, the default algorithm used in API is even faster, see the Travis logs.

goos: darwin
goarch: amd64
pkg: github.com/rvflash/fizzbuzz
BenchmarkMultiples_One-4            	20000000	        61.6 ns/op
BenchmarkMultiples_Two-4            	20000000	        59.3 ns/op
BenchmarkMultiples_Three-4          	30000000	        51.7 ns/op
BenchmarkMultiples_Bulk20-4         	 2000000	       719 ns/op
BenchmarkMultiples_BulkTwo20-4      	 2000000	       749 ns/op
BenchmarkMultiples_BulkThree20-4    	 2000000	       835 ns/op
BenchmarkMultiples_Bulk100-4        	  500000	      3422 ns/op
BenchmarkMultiples_BulkTwo100-4     	  500000	      3538 ns/op
BenchmarkMultiples_BulkThree100-4   	  300000	      3950 ns/op
PASS
ok  	github.com/rvflash/fizzbuzz	15.906s