Ethen Bommaiah
1. Would called-by trees be useful in understanding large pieces of software ?
2. How do we evaluate SV sytems for debugging ? Have there any emphirical studies to show that these systems have resulted in improved productivity ?
1. What kind of problems/errors did they use when they compare the performance between using VIPS and Dbxtool? I feel that even if programs with linked list structures sometimes it is easier and faster to have the debugger just get to exactly where it fails instead of browsing through all the structures with the human eye.
2. What practical applications do systems like FIELD have? Besides aiding in debugging, will it be helpful in design?
1. FIELD uses a relational database to provide query support on syntax and semantics and 3D graphics in the display of analyses. Do these processes add to the time complexity of the compilation process to the point that productivity is smaller ?
2. The call graph displays appears to work well with non-recursive programs, but does not address wither recursion or multithreading or forking. Does this not diminish the value of the FIELD system in a commercial environment ?
3. In the close of chapter 19, Lieberman and Fry suggest that language and compiler designers provide more "introspective features" in their designs, but does this also complicate the development process by having to do some heuristic analysis code during compilation ?
1. ZStep95 system tries to be exclusive in terms of real-time tracking and history keeping. I would be interested to see how it works out and whether it is practically useful, in particular, how they will handle the tradeoff and set up the appropriate filter for that ?
2. In the extra paper, they talk about selective display and rapid redrawing, which is a good idea, especially for real-time debugging purpose. However, it would be nicer that their node representation can be improved, which is a little bit hard to follow right now.
1. It would be very nice to have a debugger tool to monitor the program execution behavior in many aspects such as multiple views and graphic viz, and therefor to localize and locate the bugs. However, the trade-off is also apparent, such as storage and speed issues and even portability issues. What would be the major concerns when designing and developing those viz tools?
2. Is there any such debugging tool implemented to be able to suggest a user some most common mistakes undetected by the compiler?
Would the FIELD system benefit from use of a palette describing what the various colors, crosshatchings, etc. meant?
Is the 3-D encoding of attributes in FIELD confusing?