Talk 1: how do you handle the inlining of recursive functions, and loop conditions that depend on a symbolic variable?
Talk 1: How is the budget expressed? Is it by time, or work? Does the budget account for calls to Z3?
Talk 2: TAJS was mentioned once as fast a tool, and once as an accurate (but slow) one on the summary slide. Intended as different configurations of same tool?
Talk 3: Why have you choose the Ocaml language?
Ward, Zeller's first name is Andreas :)
Talk 1: platform-specific analyses can also be done at the LLVM level and debug info at some point can link back to sources, so clang SA also use these infos?
Talk 2: did you also look at tools for other dybamic languages, like Python?
Talk 3: Does Frama-C support SARIF? https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=sarif
Talk 2: Why did you use only edges for evaluation? Did you measure the impact of your manual changes to these tools?
Talk 4: after splitting the commits, doesn't it also make sense to consider merging the separate parts of consecutive commits?
Talk4: Did you deal with multiple commits for the same reported bug?
How to know if '__restrict__' should or not be added
You talk only of source code, but what, for example, about associated documentation, or is that not relevant for your application area?
Talk 4: Possibly relevant S. Zhou et al. Identifying Features in Forks ICSE 2018 https://www.cs.cmu.edu/~shuruiz/paper/INFOX_ICSE2018.pdf
Talk 1: Why did you use only edges for evaluation? Did you measure the impact of your manual changes to these tools?
Talk 2: could you explain the impact of the underlying technique on the results for each tool? Which technique should future tool builders incorporate?