Code reviews, statistics-based testing, and virtualization were hot topics at ESC Boston this year.
As someone who produces material for the online world, I sometimes under-estimate the importance of conferences and the chance meetings that occur at them. While at Embedded System Conference Boston as a presenter, part-time interviewer, assistant teardown analyst, and general TechOnline booth support, I have had a chance to participate in many informal discussions that truly make this conference a unique experience. As the Embedded System Conference series (ESC Silicon Valley, ESC Boston, ESC India, and ESC UK) are top conferences in the embedded space, many of the industries top minds are here, both presenting and attending.
Hosting a shop talk Monday morning, there was a good discussion on both hardware and software testing that hopefully benefitted most there. But as an example of how unexpected discussions can be quite valuable, a participant posed a question--how do people review their code? Another participant gave an extremely good example of how their company used the Code Collaborator code review tool from Smart Bear software. This raised the question—would using the software eliminate the need for meetings? Is there still a value with meetings even though the software can do everything? Or is there a way that the code review software can actually be used during a meeting? Technology is changing the way people work, including the very software developers that create new technologies.
Another interesting conversation ensued while waiting in the speaker ready room. A speaker from the adjoining Dr. Dobbs conference on Software Best Practices, Christian Gross discussed methods of statistics-based software testing methodologies that is just starting to see some use in the financial sector. The randomly assembled group had a quick discussion of whether these statistical based methods could ever find there way into the embedded world. The real-time, sometimes safety critical, decisions that need to be made inn embedded systems need to be critically tested for certain instances. Maybe in the future a combination of both methods could be used for embedded system testing.