Brief notes on this Webinar: presented by Jim Holmes and Dave Haeffner, and hosted by Telerik
---------------------
This Webinar caught my eye because I'm working my way through co-presenter Dave Haeffner's Ruby-based version of "The Selenium Guidebook". (A post or two to follow on that, I'm sure.)
This short webinar turned out to be a general "primer" for teams/organisations looking to adopt automation and didn't specifically discuss tools or techniques. But it was no less interesting for that.
I understand that it accompanies a series of blog posts on the topic - see the link right at the end of this post - but I haven't read those as yet.
It was nice that, despite being organised by tool vendor Telerik, the webinar did not push their automation tool. And Holmes and Haeffner are definitely not naive enough to believe the "automate everything" myths that some tool vendors seem to push. (Though they didn't specifically refer to it, I suspect Holmes and Haeffner get the "Testing vs Checking" thing too.)
The following are my notes from the webinar and don't necessarily reflect direct quotes from Holmes or Haeffner.
Why automate?
- Have a reason for your team to adopt automation, not just because everyone else is supposedly doing it. Question if it really adds value in your context.
- Use automation to solve business problems, not to cut the size of your testing team.
- Have the right expectations of automation. Organisations who expect to automate everything or to eliminate manual testing are misguided.
- Automate the key, happy path stuff.
Implications of automation for testers in the team
- Automation should enable people to be re-purposed and time to be re-allocated rather than removing the former and reducing the latter. It should be a way to free up testing resource to do things computers can't do. Enables a move from repetitive scripted or regression checking to Exploratory Testing, questioning the product.
- For testers, automation needn't be intimidating. Good testers with the right mindset can utilise automation without the need to try and become developers instead. They can still have focus on the thinking skills intrinsic to rewarding testing.
- Encourage them that automation code is not as intimidating as application code. Only a small subset of programming knowledge is necessary to get to grips with automating. In his Selenium book Haeffner talks about just 8 key concepts that he thinks are necessary for test automation. Indeed, the most successful implementations have developers helping the testers.
Implementation
- Holmes and Haeffner suggest running a time-boxed pilot of any automation implementation. This is necessary to adapt processes and set internal expectations as necessary.
- There is a danger of building badly-implemented automation that the team doesn't trust. Avoid this with good abstraction, and robust tests. Use tools, techniques (eg. continuous integration) which make automation visible and accessible to the whole team. It shouldn't be running in isolation on someone's machine. Aim for constant feedback. Learn from failure and advertise success.
- In Dave Haeffner's view, good automated checks in a suite:
- are concise and specific. Checking one, clearly understood thing.
- all run independently of each other
- are descriptive
A recording of the full webinar, and some Q&A that they didn't have time for, can be found here.
No comments:
Post a Comment