Difference between revisions of "Test Tools The Next Generation"

From CitconWiki
Jump to: navigation, search
(Formatting)
m (Formatting.)
 
Line 47: Line 47:
  
 
* What about something that enables getting end users involved?
 
* What about something that enables getting end users involved?
  - Session recording?
+
** Session recording?
  - Prototyping?
+
** Prototyping?
  - Exploration toolkits
+
** Exploration toolkits
  - Paper prototyping
+
** Paper prototyping
  - Goal: Get feedback from the customer as soon as possible to guide your development
+
** Goal: Get feedback from the customer as soon as possible to guide your development
  
 
* Mechanical Turk, how much can you control it?
 
* Mechanical Turk, how much can you control it?
  
*****
+
 
PJ
+
PJ Proposing "AI Testing", using Machine Learning because the greatest cost in testing is maintaining a legacy test suite.
*****
+
 
Proposing "AI Testing", using Machine Learning
+
because the greatest cost in testing is maintaining a legacy test suite.
+
*****
+
  
 
Mark Streibeck from Google on Testing
 
Mark Streibeck from Google on Testing
- If the test fails, and you have to change the test, then it is a BAD test
+
* If the test fails, and you have to change the test, then it is a BAD test
- If the test fails and you change the code, the it is a GOOD test
+
* If the test fails and you change the code, the it is a GOOD test
  
 
Microsoft Reverse Code Coverage tool?
 
Microsoft Reverse Code Coverage tool?

Latest revision as of 00:12, 12 September 2015

CITCON 2015 Europe in Helsinki

Session 1 - Auditorium

Testing Tools of the Future - Proposed by Gojko Adzic


Given how cheap resources (machines, people, etc), what can we do now that we could not do before.

Existing Examples: https://www.usertesting.com/

https://aws.amazon.com/device-farm/

Reminder of Jason Huggins's

  • Why not just do production monitoring, like canary testing?
    • It's a matter of risk

Third interesting trend: Webkits really grown up

Gojko uses Phantom for DOM testing all the time.

Wraith, from BBC, is a testing tool that runs the test in lots of browsers and then diff the visuals

Gatling, another tool that was very active

James Shore released Quixote, CSS Unit Testing, tool

Gauge was released by ThoughtWorks, as replacement for Twist


Classes of Problems:


  • Fragmentation
    • Issues of testing across multiple devices
  • Resilience
    • Issues related to the ephemeral nature of the infrastructure
  • Costs
    • of Execution
    • of Maintenance
  • Infrastructure
    • How to test the infrastructure


  • What about something that enables getting end users involved?
    • Session recording?
    • Prototyping?
    • Exploration toolkits
    • Paper prototyping
    • Goal: Get feedback from the customer as soon as possible to guide your development
  • Mechanical Turk, how much can you control it?


PJ Proposing "AI Testing", using Machine Learning because the greatest cost in testing is maintaining a legacy test suite.


Mark Streibeck from Google on Testing

  • If the test fails, and you have to change the test, then it is a BAD test
  • If the test fails and you change the code, the it is a GOOD test

Microsoft Reverse Code Coverage tool? To speed up feedback, by running

A/B Testing can be more difficult in Finance applications

One Company that Gojko knows of wrote a test, but as soon as it passes, they deleted the test

False positives, test designs

Is a "test suite" a _tall_ factory?

Pascal community used mathematical modeling

Google captures a list of system attributes/capabilities

Unit testing: if the unit test is hard to write, then that indicates a code smell


References

=====

Quixote, CSS Unit Testing Galen Framework Wraith, from BBC Device Farm from AWS - https://aws.amazon.com/device-farm/ UserTesting.com Gauge Concordian Applitools Visual Review Domreactor Hotjar Text Test - for testing log files