The Google LTAC (London Testing Automated Conference) drew to a close on Friday, with a number of interesting presentations and opinions voiced over the last couple of days. Thanks are due to Google's Allen Hutchison, as well as the remainder of the team at Google who pulled it all together.
The conference itself, at Google's London headquarters, was typically, well ... Google. There was a scrolling sequence of popular searches at the reception, as well a Google Earth demo that was bungy-jumping from what appeared to be one capital city after another (perhaps Google's offices?). The cafeteria (which we'd occupied for the conference) has a pool table, as well as an array of free drinks, confectionery and donuts. And, in typical Google style, the lighting illuminating the stage was blue/red/yellow/blue/green/red, as can be seen in some of the wide-angle flickr photos (such as this one). There's also an open balcony which is open to the London elements; but overall, a very open space.
The other thing which is becoming more common is the use of virtualisation technologies (like VMWare/Xen etc.) for testing purposes. I especially liked the demo of the MacBook Pro running many concurrent VMs and being able to record their output via VNC to SWF (even if the demo didn't work properly at the time; it shows the potential). Several other uses of VMWare are becoming more frequent; for example, using VMWare images to run real applications is one thing being explored by others (e.g. Amazon's EC2) as well as shipping apps like Google's search appliance.
Literate programming -- and in particular, literate functional testing -- seems like a very neat idea (much like Apple's AppleScript). Unfortunately, also like AppleScript, some programmers really don't get how to write this kind of literate programs, given that they are more used to procedural programming languages. It was neat seeing a literate functional program tests being stripped of its punctuation and being relatively readable; though similar results can be done (at a higher level) with tools like FitNesse, which looks like it's gaining a lot of ground (and of course, FitLibrary). I suspect that the tests created by the Fit systems are likely to be more usable than the literate programs, because declarative programs are probably easier to write (and to know that you've got right) than the literate programs.
The conference evening was held at The Camel, just down the road from Google. Unfortunately, there wasn't much space in the pub itself (and the music was too loud) to have any kind of sensible conversation; the people who did stay ended up spilling onto the street outside to be heard. The only other downside was the fact that there weren't any (easily accessible) power points to recharge laptops; though fortunately I found a hole-in-the-floor that I could use to keep the blog going if not real-time, then at least the same day. Unfortunately, we ended up having to time-share the socket (fortunately; all of us sitting had PowerBooks, so we shared the same charger). For the second day, I took in a multi-socket adapter so that there was less in-fighting :-)
I was somewhat surprised there wasn't any more mention of TestNG at the conference; perhaps it's just that it's already a lingua-franca of testing. But then the topics were many and varied, and there were more discussions about underlying principles than specifically tools themselves. Still, if you've not read about TestNG before, then have a look; it's got a lot of features that JUnit's been missing out for a while (though JUnit4 adds some of them).
The videos and presentations for the conference are bound to be made available in the near future; when they do, I'll update this (and other entries) on the subject.