index data structures touched?

Could it make sense for semcode to also index every data structure touched by each function?

Then it would be possible to quickly find "every call chain that touches data structure X", which could speed up code audits.

The downside is that the semcode database would likely be much larger, and potentially also much slower to build, so this may not be something people want to enable by default?

This is just an idea at this point, with no real plans on how to accomplish it, or a good insight into whether or not it should be done. Feedback is very welcome.