Gareth is a framework that allows you to do automatic validation of your business goals that are part of a user story.
We believe that in the trend of delivering software fast using continuous delivery and DevOps that the delivery of software is mainly focussed on delivering functionality and not on achieving business goals. We want to create and delivery tools that helps you to automate and validate the value of your software from a business perspective.
We want to offer a DSL that allows you to describe your business goals, this DSL can automatically be parsed and glued to code. This code run in production and allows you to set baselines and validate your goals automatically.
We want to deliver these parts:
- A dashboard that gives you a overview of your experiments (goals you want to achieve)
- A DSL
- Parsers for this DSL in different languages. (currently Scala)
- Agility: detect and act upon unused code
- Act upon success and failure automatically
- Gain maturity of user-stories
A example of the framework given below:
A experiment as part of a user story, in this experiment the the registrations are expected to increase with 50 in one week and with 300 within a month. (A experiment with two assumptions)
Experiment: Increased registrations
Baseline: The current amount of registrations is 200
Assumption: The amount of registrations is increased with 50
Time: In one week
Baseline: The current amount of registrations is 200
Assumption: The amount of registrations is increased with 300
Time: In one month
Failure: Send a e-mail
Baseline: The current amount of registrations is 200
Assumption: The amount of registrations is increased with 3000
Time: In one year
Success: Sent cake to developers
In this example description above a product owner for instance describes what the functionality is expected to do in production.
The following DSL keywords are available.
The title of the experiment, a meaningful text for other to understand
The Baseline: keyword, is part of a assumption block and describes to the first measurement you want to do at the beginning.
The Assumption: keyword, is part of a assumption block and describes the measurement you want to do after the time defined in the Time: keyword.
The Time: keyword, is part of a assumption block and describes in how much time the assumption must be validated.
The optional Success: keyword, is part of the assumption block and describes to code when the assumption is met.
The optional Failure: keyword, is part of the assumption block and describes to code when the assumption is not met.
The definition is actual glue code that is written by the programmers to validate the assumptions. This code can for example check the count of registrations in a database or check log files. There is no limit because the definitions allow you to connect to any framework or any other services to get your data.
A example definition written in scala
import nl.codecentric.assumption.dsl.core.definition.BaseDefinition
import org.json4s.native.JsonMethods._
import scala.concurrent.duration._
import scalaj.http.{Http, HttpResponse}
/**
* Created by hylke on 20/07/15.
*/
class RegistrationDefinition extends BaseDefinition {
implicit lazy val formats = org.json4s.DefaultFormats
"Get the current count of registrations" baseline (() => {
val baselineCount = getRegistrationCount()
println("Baseline count is", baselineCount)
})
"The amount of registrations is increased with 50" assumes (() => {
val assumeCount = getRegistrationCount()
println("Assume count is", assumeCount)
})
"In one week" time (1 minutes)
"Send a cake to the development team" success (() => {
println("Sending cake")
})
"Send a message to the product owner" failure (() => {
println("Having a firm talk with the product owner")
})
def getRegistrationCount(): Integer = {
val response: HttpResponse[String] = Http("http://33.33.33.60:8888/count")
.timeout(connTimeoutMs = 1000, readTimeoutMs = 5000)
.asString
val json = parse(response.body);
(json \ "count").extract[Integer]
}
}