Rant on Login Screen examples

If you are demonstrating testing technologies or testing examples around RPA, ML, Selenium and so on – Please: DO NOT USE A LOGIN FORM!

The test scenarios I usually deal with are not this… mundane. While a few testers probably still have to build login forms from scratch, a login feature is a commodity by now. Use OAuth for public facing sites and Active Directory Federation internally in the organisation. Really – there’s no need to reinvent the wheel. To the end user and even the Product Owner logging in is just a stepping stone. 

I just want to log in, and then I’m done for the day

Said no user ever*

Showing that you can train an AI network or other framework to login might solve a tedius testing task, but is usually not the thing I’m after. When a user is logged in, they are there to solve something, to process something, to do something – to engage with something. And this is where the best tests are heading too – this is the tests that adds value to the business and tells something about the product. 

For instance: In one project I did, we disabled the login entirely to make the CI/CD run feature testing. The plain login screen was temporary anyways, as the solution authentication would be based on certificates. We never spend much time on it, neither on total coverage.

Less combinations – More Real Life Scenarios

So what can you use as an example instead of login boxes and combinatoric bar stories? How about anonymizing the latest test you had to do on your latest live-action testing project? This will tell me about your challenges, your business domain and when the last time was – among other things.

Let me start, as of writing the latest test case I touched (same day as writing this) was for a new public data registration project. The tester and end user “subject matter expert” was testing the data registration form from both a GUI and web-services perspective. Export the entered data form as XML and import it again via web-services to see the consistency.

Could that be tool supported – maybe. The knowledge about the system is not very explicit. It’s a bit complicated, actually. Could it be trained by an ML/AI system – I doubt it. There is no global training set for this class of systems – we have the old version but are adding new things. 

If you are (a tool vendor) demonstrating something – do try to understand the context and problems your customers are trying to solve first. Ask them what their latest tests were and where the challenges are. If it’s a login box you’re good to go, but I doubt it. 

*: Unless you are somehow measured on “logging in”. For instance to claim unemployment benefit, login to job site daily…Snap! OTOH shallow measure.

http://rebrick.lego.com
Advertisements

The Guilty Tester Episode 7 – – With A Little Help From New Friends

Do you ever feel guilty for not meeting the standards set by others in the Software Testing community? You’re in the right place then.

In this episode I talk to Dave (@theguiltytester). We discuss traditions, open questions and how to work within contracts which are specifically requesting traditional test practices based on large numbers of test cases. 

Listen and read all about it here: http://theguiltytester.libsyn.com/the-guilty-tester-episode-4-jesper-ottosen-with-a-little-help-from-new-friends 

Some of the blog posts mentioned are: 

New friends - the subject matter experts of all trades
New friends – the subject matter experts of all trades

You don’t have to be a boss to be a leader

It’s really that simple, yet awesomely profound. And a typical Gerald Weinberg quote, like my other favorites of his points:

  • No matter how it looks at first, it’s always a people problem (The second law of consulting)
  • You’ll never accomplish anything if you care who gets the credit
  • Things are the way they are because they got that way
  • Quality is value to some person

Regarding the last quote; which was later extended with “who matters, at some time” by Bach, Bolton. Once I had an argument about how to deliver quality. The other side held towards IEEE definition of delivering the expected. But even when he did, he failed to see that the unmeasured and irrational parts affected the value to the customer. I agree completely with The Cowboy Tester that knowing works of Weinberg is a measure of seriousness.

Weinberg worked not only with testing, but among other things also consulting and organisational change management. I did not know that when reading “Making Sense of Change Management” (Cameron & Green 2012). I literally jumped up and started laughing while reading the quite serious elaborations to the Satir Change model – the authors found that Quality Software Management: Anticipating Change (1997) is a “masterly book on change, but with a title that might not appeal to everyone“. It might not appeal to change scholars, but definitely appealed indirectly to a lot of people in testing.

Recently (August 2018) Jerry died aged 84. Not a boss – yet a leader.

Could Modern Testing work in the enterprise

So far I have mostly thought that “Modern Testing” of the A/B testing podcast would never work in an enterprise context. But it seems some tools and existing approaches in the enterprises already fits well with the ideas of the concept. 

The enterprise is all the privately owned companies that usually manufacture (non-IT) things –  for either the consumer or other businesses. The enterprise sell and produce tangible products like windmills, power tools, dairy consumer products etc. The interest with regards to IT for the enterprise is that it just works, and supports business processes around order setup, order tracking and invoicing – and many other moving parts.

While I have heard of some organisations that have successfully implemented some agile and SaFe methods (in banking), the enterprise have a hard time to change mode of operations, as it usually comes down to actual production of things, logistics and hierarchies of command-and-control … and culture, most of all culture.

 

@  via @HelenLisowski

 

Some enterprises change towards being learning organisations, but still treat their IT in general as low-value and an annoying cost. It seems the IT departments and IT contractors have a challenge in talking about what they do to achieve the right quality for the businesses….

 

 

Que: The Modern testing mission on “Accelerate the achievement of Shippable Quality”. While MT is mostly a concept around transition of testing activities, it seems the concept applies to IT delivery teams in general. MT has 7 principles and some of these are:

5. We believe that the customer is the only one capable to judge and evaluate the quality of our product

Most enterprise projects I know off around implementing SAP, MS dynamics 365, EPIC hospital solutions etc, have a large reliance on end-user testing and UAT. Often there is no professional testers involved, as the best tester is the business experts themselves. Interestingly the principle #5 fit’s well with existing practices from the UAT space.

Another interesting MT principle is #6 around data analysis of actual customer usage. This requires some totally different tooling for the tester, than previously generally available (…besides shifting-right perhaps…).

6. We use data extensively to deeply understand customer usage and then close the gaps between product hypotheses and business impact.

Yet recently I was investigating Panaya Autonomous Testing for SAP. One thing I realized is that what the tool collects real user usage of SAP and then provides the ability to balance the testing activities based on that analysis. It is interesting to see a commercial test management product for the enterprises following new trends like the “modern tester”.

While it’s interesting how some of the concepts of modern testing are reflected in testing in the enterprise – and vise versa –  the challenge remains for both the tools and concepts to be applied and accepted in more organisations. 

It might not fit everywhere, but it might be a good fit in more places than you think it would. 

 

Recruit for Curiosity

Recruiting for testing roles these days should be mostly about curiosity, problem-solving and less about productivity and text book knowledge. Recruit for right brain skills – not so much operational process jockeys. 

Recently at UKSTAR 2018 Simon Prior talked about his investigation into University programs and their rare courses in testing. This lead to his twitter discussion under the tag: #makeAtester where the top responses of skills required was curiosity. Quite in line with the State of Testing Survey 2017 that lists key communication skills as most looked for when hiring for testing roles. Both surveys establish that testers are knowledge works.

IMG_3614

Similarly HFresearch have compiled an analysis that even on a management level the trend is to hire for creative thinkers over “operational experts that improve business performance and productivity”.  Talent focus should be on right brain thinkers over – The wonks who spend all day staring at spreadsheets, focused on execution “left-brained” activities are less in demand .

But where do we find curiosity training?

If that is the skills we are looking for perhaps we should stop looking at university programs in computer science or engineering, when we want to recruit testers. I have a computer science master degree, and that was really theoretical and while it somewhat focused on problem solving, the lesson was rarely about thinking outside the box.

I think I would rather higher with business domain skills and train testing theory, than hire a process jockey with no experiences in besides textbook examples. That’s also how I came into testing myself, practical activities first – formal training later.

Perhaps it’s not as such important to have an university degree to get into testing. Though it helps 🙂 A diverse background is important, I know of librarians, laboratory technicians and humanities majors that bring good competencies to the testing field.

Finding one higher education that focuses on building curiosity, whole picture thinkers is hard – perhaps dungeons and dragons, as also discussed at the conference?

Shifting is more than Shift Left

Links on the topic: Shifting is more than Shift Left as presented at the OnlineTestConf Fall 2017.

My own writings:

by @KatjaBudnikov #katjasays
by @KatjaBudnikov #katjasays

On Medium regarding Testing, AI, ML etc

I’m writing on Medium regarding Testing, Artificial Intelligence, Machine Learning etc:

More to testing than AI and ML can solve

Artificial Intelligence (AI) and Machine Learning (ML) can perhaps solve some testing challenges, but not all testing. The testing vs. checking debateand all the shift-left of checking, have revealed that some of testing is about critical thinking and some…

 See also

Similar posts: Testing is wicked, Test all the things,

mad-brain