AFLplusplus / AFLplusplus/LibAFL
Any reason for not having a common interface between feedbacks and observers?
- Dominant language
- Rust
- Stars
- 2.6k
- Forks
- 481
- Avg merge
- 2d 30m
- Merged PRs (30d)
- 16
Description
I see 4 types of relations between the feedbacks and observers
1. Feedbacks that accept any observers (e.g. Diff)
2. Feedbacks that accept observers of specific traits (e.g. Map, withHash, etc.)
3. Feedbacks that accept observers of specific types (e.g. list, std, etc.)
4. Feedbacks that do not accept any observers (e.g. const, crash, etc.)
That is to say that there is no solid interface between the feedbacks and the observers, also as per the paper.
Have you thought about standardizing this interface, or are there any super dauting challenges?
The way I see it, the only visible difference amongst Observers is their output.
So is it not a better option to have a standardized interface that outputs a generic instead?
Maybe I am not seeing something you guys have seen before. If so, would love a conversation on it.
Contributor guide
Assessment
This issue has not been assessed yet.