From my experience I tend to say AnthonyGA has a point. The biggest airplane I ever flew yet is the Challenger 300, but that airplane has (had?) enough software generated problems and also problems that origin from the analog / digital interfaces. For example, we had erratic engine behaviour which was not reproducable, in the end we replaced the throttle quadrant (comes a s a complete unit) and thing worked as advertised again. The test and measurements where all within limits. I nowadays fly the Cessna C680 and on our airplane we sometimes donīt get the HP bleed from one side when anitIce is on. MX have checked everything and the condition is not reproduceable at will, it just happens sometimes. I found out that we could get it to work with changing fadec channels, however P&W and Cessna say its not a FADEC (aka software) problem. What is it then I ask? ADCs Rat probes etc have been checked....
And how do we deal with problems we can reset by cont/alt/del aka cycling the power? Hold item, MEL or just shrug shoulders? A wide field that needs a bit more attention IMO.
Regarding the certification of software, Iīm not involved in that, but I ask myself how can they get stuff certified that acts sometimes really strangely, like the Honeywell Epic avionic suite installed in the C680...You are not allowed to use a lot of functions that originally have been okay to use and we get them back in little steps. The magical phase 5 software that is supposed to cure all the probs is delayed time and again, so methinks its a lot harder to find out and rule out software related problems than hardware.
Sorry for the OT.