[Wtr-general] A general question about the role of scripted testswithin the responsibilities of QA.

[Jonathan Kohl]

>At what point do these scripts allieviate some of the QA engineer's manual
testing burden?
Until machines are intelligent, they can't really do testing at all. Testing
to me is assessing risks, asking questions about software, using a guide to
show me whether something is wrong, using techniques and tools to help ask
questions of the software, and analyze results. Automated testing tools
can't do any of that, but as a tester I can use them to help increase my
efficiency. What task is a computer better at, or faster at, and how can I
use the tool to make my testing work more efficient? So to me the manual
testing is essential to complement the automated efforts. Automated tests
are great for quick passes of certain kinds of tasks, and are nice for
change detection. 
 
They can't replace a human, but people can use them for more efficiency.
Instead of doing a smoke test suite manually that might take hours or days,
an automated smoke "change detection" suite can give me a rough
approximation of changes in minutes or hours. Then as a tester I can focus
on areas of risk, and spot check trouble spots. Efficiency is key to me, not
trying to replace manual testing work. The former is achievable, the latter
isn't going to happen. Over reliance on regression tests to do testing for
us means that obvious problems a tester would spot immediately go out the
door. The human eye can pattern match better than a computer can, and the
mind behind the tools can investigate, use inference and track down tricky
problems. many times when we are running a manual test, we discover a
problem that is not related to the test. Many of the high impact bugs I've
come across were something we stumbled upon when testing other things,
investigated and fixed. A computer isn't curious, aware and can't
investigate suspicious behavior like a tester can. I find those qualities
valuable.
 
My $0.02
 
-Jonathan
 


  _____  

From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Michael Kelly
Sent: August 18, 2005 10:21 AM
To: wtr-general at rubyforge.org
Subject: [Wtr-general] A general question about the role of scripted
testswithin the responsibilities of QA.


My developer chums and I use Watir as part of our test-driven development.
As such, we're giddy to be able to finally write unit tests for the UI.  But
the presence of these developer written UI test scripts raises questions
about what impact, if any, they will have on what our QA engineer will focus
on.  It's tempting to suggest that there is a whole set of functionality
that they simply don't have to test manually anymore.  But clearly this is
just wrong, wrong, wrong.  The developers will only write tests for the
things they think of, and we all know that developers tend to be optimistic
about their code.  In addition, the unit tests themselves can have flaws
that cause a test to pass when the code is not, in fact, functioning
properly.
 
So, it seems that we still need the QA engineer to do a full manual QA pass
on the software.  At what point do these scripts allieviate some of the QA
engineer's manual testing burden?
 
Thanks for your thoughts,
 
-=michael=-

-- 
Michael Kelly 
Sr. Software Engineer 
Eleven Wireless Inc. - The Possibilities are Wireless 
http://www.elevenwireless.com <http://www.elevenwireless.com/>  

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/wtr-general/attachments/20050818/10c607aa/attachment.html