Bradley M. Kuhn
Mon, 1 Feb 1999 00:49:57 -0500
Thus spoke Kragen Sitaker:
> Other things I've been thinking about: it would be *really* helpful to
> be able to click on an identifier in the source and immediately be
> taken to the definition of what it refers to. Since the compiler and
> linker have to figure this out anyway, it doesn't seem like it would be
> too much to ask for them to write it down so we can see it.
> Such a database would be very helpful in answering questions like these:
> - What is the set of functions that could get called, directly or
> indirectly, from this one?
> - What is the set of places where this data I'm writing could be read?
> - What is the set of places where this data I'm reading could have been
> Given a database of what-identifier-refers-to-what-definition, these
> questions would be trivial to give reasonable answers to. (Of course,
> sophisticated analysis could reduce the size of these sets somewhat,
> and wild pointers can violate anything.)
> What other questions would you like to ask a database like this?
I would look at Keith Gallagher's work on program slicing. Start from
www.cs.loyola.edu/~kbg, and move on from there.
He attempts to ask these kinds of questions and answer them. They are less
trivial then you think...
- email@example.com - Bradley M. Kuhn - firstname.lastname@example.org -