zaks.anna added a comment. I agree that we should not print the values of all variables. The users will be overwhelmed with the huge amount of information. It is very valuable to highlight just the right information. (I believe even the current diagnostics often produce too much info and highlighting the more important notes would be very valuable.) However, the examples you presented are **very compelling**. Are there ways we could highlight the same information without printing values of all variables?
For example, for the overflow case, we could experiment with printing notes along the path at the locations a variable overflows. This would be very valuable for the array overflow alpha checker, which often flags the bugs that only occur if an integer value along the path overflows. I am not sure how noisy this approach will be. If it is too noisy, we could refine this further. For the uninitialized variable example, we could have some pattern-matching logic that would check if the expression is an array element and if so print the value of the index. (Another problem with variables that are failed to be initialized in a function is that we do not display the path on which the variable is not initialized, making it hard to understand the reports. Though, that is a separate problem.) https://reviews.llvm.org/D34508 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
