Squidlord:
For a time in the 90s it seemed to be a de facto policy for safety-critical avionics software but I wouldn't know whether it was ever formally required.
Curious to know when in the 90s you are referring to here. We have been told that those years were the "Golden Age" of Military Airworthiness Provision by someone, who was told that by someone else....Given that this was the decade that produced this code that was declared positively dangerous by BD, an MAR fuel system that was equally so, an IFF system that did not fail safe, and the continual pleas for a tactical transport's fuel tanks to be given minimal battle protection at least ignored, oh, as well as the replacing of acl's with strobe lights in contravention to regulatory procedures that should have been followed, I wouldn't want to be around when it wasn't a Golden Age! BTW all those airworthiness shortcomings were involved in Airworthiness Related Fatal Military Air Accidents that claimed a total of 62 lives. It seems to me that de facto policy did not lead to de facto observance of the Regulations, but that is something that you have said is not strictly necessary, isn't it?
I disagree with Chugalug2 when (s)he suggests that if regulation is not followed, it necessarily means the software is unairworthy.