Thursday, May 08, 2008

StarEast - Update for Wed/Thursday

OK, so I didn't get the Wednesday post written on Wednesday. I forgot how the evenings get busy here, too.

So, here goes. I liked James Whittaker's opening keynote session on Wednesday. It was a different topic than in the brochure, but that's cool.

He made a very important point that our future will involve a lot more code than it does today - and it's going to control even more critical functions than today. Therefore, it needs to work right all the time. I thought the session had some major implications:

1) He's talking about defect-free computing (not just software, but hardware, data and everything else involved in getting correct results.). History has shown this has been an elusive, if not impossible, effort to achieve - largely due to the complexity of the code. Does that mean we should give up? By no means. But, zero defects requires some very rigorous methods and ways to build-in quality, not test it in, which was one of James' points.

2) He said testers need to be less like Lewis and Clark in terms of their test approach. OK, I buy that. Lewis and Clark were explorers. I take that to mean that exploratory testing has some limitations and we need a better way to test. It's interesting, though, that exploratory testing is a very popular test approach and has gained a notable following. Is this a blow to exploratory testing - or simply an admonition that other methods are needed? I think it's the latter, but I find the remark very, very, interesting.

3) James said that testers need more insight into the code. Black box testing is inherently inefficient because you do a lot of poking around. I agree. So, the question is how do we get the development tools to the point that they also contain great diagnostics?

4) Software should be so good that testing is no longer needed. I only wish. I doubt that will ever happen due to the human aspect of software development and usage. About 20 years ago, the folks at SEI didn't include testing in the CMM because they felt if you had a good enough process, the software should be defect-free (or close to it). That never happened and I doubt it will in the future, either. So if you are a tester I wouldn't worry about your job security, at least as a profession. By the way, 20 years ago, people were predicting that coders would not be writing code in the future. While a great deal of code is generated by tools such as Microsoft's Visual Studio, there is still a whole lot of manual coding going on!

I don't know if we'll see the digital future James portrayed in his talk, but I do agree more and more things will be software-driven and the criticality of the applications will also be higher. Think of the cars that will be driving themselves. Heck, there have been cases already where Volvos have quit at highway speeds due to software failures. Along with the cool technology comes problems that aren't so cool.

I liked his session a lot and hope it resonates as a call to testers and developers to take software quality to a higher level.

The second keynote address was by Elizabeth Hendrickson, who spoke about her experiences as a tester on an Extreme Programming team. It was a great talk as well and gave people a good perspective of what an agile tester's work day is like. She addressed issues like requirements in agile, which I thought was great.

Throughout the day, there were a wide variety of track sessions which were well attended. Not every session had a great speaker, but sometimes the content is great so you stay. Sometimes the session as a whole just doesn't do it for you, so you can move to another one.

I spoke at 3:00 on the topic "Testing Disasters and Turnarounds". Thanks to everyone one who attended. I have posted my updated slides here:

The products and services expo was good - very well attended. However, it seems like the numbers of tool vendors was smaller this year.

To top the day off, there was a casino night with several really cool prizes, none of which I won.


Bahrat Mediratta and Antoine Picard of Google gave an encore presentation of their keynote from StarWest called, Testing in the Toilets. It's a neat story about how a simple act of posting articles they write about testing in the one place everyone goes - the restroom. The results were interesting and it's a great story about how they changed the culture at Google in terms of test awareness.

I had a book signing and other duties today, so I only got to attend two track sessions. I thought the one by Gerard Meszaros (author of "X Unit Patterns) was very interesting on building record/playback automation into your applications. It's an interesting alternative to commercial tools and opens some doors on some creative test automation that solves many of the problems seen in traditional test automation.

Finally, I went to John Fodeh's keynote session, "Are We Ready to Ship?", which was a nice treatment of the topic of release metrics for software. I came away with a lot of good ideas.

So, that's it. I'm heading back home early in the morning, so this is my last StarEast post this trip. I felt it was a good conference. I got to see many good friends from all over the world and that always is a good thing!


Annie said...

Thanks for the great writeup, Randy. I really enjoyed attending your talk (it's always good to know that disasters occur everywhere) and the conference overall was wonderful. I've been reading the disk you handed out on being a good test team leader and it's very thought provoking.

I also think you're right in your interpretation of James' "Lewis and Clark" metaphor, but I think that exploratory testing as a whole should also be reexamined. Lots of people want to be lazy about documentation and I feel like this is their back-door to getting away with as little as possible!

Randy Rice said...

Hi Annie,

Thanks for your comments. It was like throwing meat to Piranhas when I said I only had a dozen or so CDs left. So, if anyone missed out on getting a disk, let me know and I'll mail one to you (US mailing addresses only, please).

One very helpful addition to exploratory testing is Jon Bach's session-based testing. I thought his track session was great on this topic. You can also find info on this on the web.

To me, the take-away is that testing needs more visibility into the item being tested than pure black-box testing can deliver (e.g. gray box testing).

I agree that many of us take the path of least resistance in testing!

Thanks again for your comments, Annie!