Deterministic testing of event loops

williamedwardscoder.tumblr.com

26 points·willvarfar··3 comments

This vaguely reminds me of Microsoft's CHESS testing tool: http://research.microsoft.com/en-us/projects/chess/. An interesting research project that (IMO) didn't get its proper day in the limelight.

I want to believe things like this could be useful, since I spend most of my professional time researching and developing systems to mechanize the process of testing and flaw discovery, however reading this was sobering and depressing. symbian doesn't exist any more. that they did very intelligent things to remove bugs doesn't really matter. there are some companies that do very smart things to find and fix bugs before shipping, and they're around, and others that don't, and they're still around too.

so what's your incentive to do this, as a software organization? if it doesn't matter and it's expensive and it generates you more work that I guess doesn't matter if it's done or not... why do it?

Symbian has been superseded but when this tool was made it was going strong.

The strong charge on reliability was mostly a UIQ pride thing; everyone would say it was important but only really my team and some people in Symbian itself tried to do much about it. (Coverity was used very effectively towards the end.)

Most importantly though some goals came from the top as every phone returned to a shop by a disgruntled customer was charged back onto the product group by the support group.

Reply on news.ycombinator.com