Over time, software testing will become fully automated and embedded in the processes and technologies within the overall software development life cycle.
Thoughts on the article? I wonder if theyâll be adjusting their training courses in accordance with this document in the near futureâŚ
I donât believe testing will ever be âfully automatedâ.
Some aspects certainly can be, like API testing. I have worked with scenarios that require making sure the response code is a certain value which is a yes/no answer.
However, I donât believe we will see a time where automation is favoured over manual checks for things like accessibility and usability.
Also, what if I only want to run a test case once or twice? Iâd spend longer coding the test and then battling with flakiness than if I just tested it myself manually.
The ISTQB qualifications are controversial though in terms of whether they are useful or not so Iâd take whatever they say with a pinch of salt.
Iâm with Faith on this. We have all come across apps in our everyday lives where itâs clear that the code works 100% but that the system was never evaluated by a human being before rollout because it behaves in ways that run contrary to the ways human beings perceive, think and act.
Any system intended to be used by people needs to be tested by people.
Another +1 for Faithâs comment - I get the feeling that whoever wrote the ISTQB paper has bought into the hype about how everything can be automated and AI can be trained to create the automation. Yes, we can automate happy paths and predictable error cases, but you really need a human doing exploratory testing on top of your automation pack if you want real assurance.
The article was a though-to-read one and in all honesty, I donât like it in any way (be it a message or the writing style). They speak of âeverydayâ things like it is something futuristic.
Some of their ânewâ messages:
quality qualifiers always change
It was like that always
quality assurance should be everywhere
It is. Trust me. Even your shirt went through some QA
automate everything
Something everyone strives for already, but everyone sane knows it is not 100% achievable. They just tried to sprinkle some âAI/Machine learningâ-like terminology on it so it sounds cool.
promote testing
This might be only 10 years old message. Donât know when QA gained the popularity and lost the mantle of the hated team.
engage stakeholders
I think we have this since âManifesto for Agile Software Developmentâ
software systems are spreading amongst the general populace
Again like 10 years ago
If I did misunderstand it, I am open for any comments/advice.
Agreed on testing and âfull automationâ particularly around the classic ânon-functionalâ areas.
If ISTQB want to introduce courses on continuous testing, DevOps etc. then great. But there is a risk that the âtesting vs checkingâ argument becomes even more prevalent particularly given ISTQBâs position in the industry.
I would like to make a bold comparison: testing is like a yeast culture, there are highly bred forms in the big industry but in the wild you find undreamt-of varieties whose aspects are lost in this. So one document does not manage to bring them all together. But there are some nice inspirations. For me, I take it with me that testing resources may be also a good source as an educational tool so making testing a comprehensive util for both. What remains is the problem of implementing all this - but do not let up to make YOUR vision come true
A perfect working software for human test by machines may not be the same feelings human may have about the app.
We are two different kind and one still can not think at the foreseeable future.
Itâs still different even the robots can think. Itâs like you are telling me siri or google assistant has real human feeling when it interacts with people.
Automation is a test on rails, itâs testing something specific by looking for something specific. We can expand the number of specific things it looks for but itâs still going to be in the confines of a group that is defined.
AI/ML may expand this defined group without the need for manual intervention but I still think a manual tester is going to be able to spot things on a GUI which donât fall into that defined group, i.e. manual testers can go off the rails.