This has to be the best, industrial study comparing programming languages that I’ve seen. They’re doing it on the same system. There’s details to assess what was difficult, what got fixed the most, and so on. Summary: the developers collectively had more C experience than Ada experience, many were using Ada for first time, the Ada developers had fewer defects with gap getting wider with experience, and the Ada features cost less. A clear win [in one project] for Ada.
That conclusion also brings to mind the practice we’ve seen (eg at NASA) about using C over Ada so developer familiarity with C reduces risk. The idea was learning a new language with its gotcha’s will increase error rate in their project. It was a good hypothesis but was refuted in this particular project. Worth some studies to test it further.
Got it out of Derek Jones' references on empirical data in software engineering if anyone wants to dig out more good ones. That's a huge list of papers to go through.
That conclusion also brings to mind the practice we’ve seen (eg at NASA) about using C over Ada so developer familiarity with C reduces risk. The idea was learning a new language with its gotcha’s will increase error rate in their project. It was a good hypothesis but was refuted in this particular project. Worth some studies to test it further.
Got it out of Derek Jones' references on empirical data in software engineering if anyone wants to dig out more good ones. That's a huge list of papers to go through.
http://www.knosof.co.uk/ESEUR/ESEUR-references.html