Is software testing stuck?

James Whittaker published a blog where he wrote that Software Testing was still stuck on the 90s and that what may seem as innovation in the field is basically thrown away and discarded as unusable.

I respectfully disagree with his point, and I think that software testing has moved forward light years, at least in the last 13 years that I have been around the field to see.

I wrote more about it in my blog, but in short:
- Starting from methodologies and approaches such as ET and Model Based Testing.
- Automated testing with Keyword Driven Testing and the low level analysis currently achievable with load testing platforms.
- And maybe most importantly the growing trend recognizing software testing as a career and not so much as a thing young developers can do for a while until they learn to write code.

What do you think?
Have we really moved forward in the last 10, 15, 20 years?
Or as James writes, these "improvements" will eventually fail to stand the test of time?

▶ Reply to This

Replies to This Discussion

Permalink Reply by ShaneMacLaughlin on March 12, 2010 at 1:54pm: For me, think the most important point in James' blog was that 'testing infrastructure (will be) built into the platform'. This will have less impact on the professional tester than the test tool manufacturers, but the trend is certainly towards developing software that is inherently more testable. Presently, GUI automation is difficult because it has to wrestle with interfaces that weren't designed to suit automation and aren't expressed in terms of domain logic. Unit testing is weak because it can't handle system testing particularly well. Manual testing is inefficient when it comes to large laborious regression tests. Yes, we can combine techniques and tools to come up with solid testing strategies, but I tend to agree with James that technology hasn't delivered as much as it should have to the tester over the last few decades. In my opinion, developing software that is testable in an efficient manner is a requirement, much the same was as functionality and performance are requirements. As a developer the tools to achieve this goal are pretty thin on the ground, and if the good folks at Microsoft would care to address this problem I for one would be a happier camper.

If this is the case, it can only be good news for the professional tester, as it would help remove certain parts of the drudgery associated with testing some software, allowing more time to be focussed on the meatier aspects. I don't think software will test itself, but would like to think we could have software with robust internal record / playback and audit cababilities.; ▶ Reply to This