David Kirkby wrote: > It seems a natural assumption to me that software that generates lots > of compiler warnings from gcc is likely to be more buggy than software > which does not. My 'gut feeling' is that if a developer has taken the > trouble to write code that does not generate compiler warnings, he has > probably taken care of other parts of his software too. > > But is anyone aware of any published literature proving this fact? We > might all suspect/guess it's true, but that is not proof. > > I found this article: > > http://www.springerlink.com/content/317x276846767585/ > > "Empirical analysis on the correlation between GCC compiler warnings > and revision numbers of source files in five industrial software > projects" > > I can't actually read it without paying, but part of the abstract says > "We use such correlation to conclude that compiler warnings may be > used as an indicator for the presence of software defects in source > code. " I think it is worth noting that correlation does not equal causation. Perhaps compiler warnings are a symptom of lack of attention to detail. That fits with your premise that a lack of warnings may signify that the developer has been more thorough. However, I think we should keep in mind that this is all in the realm of conjecture.
