Details
- Reviewers
eugenis
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Another option (probably better, WDYT?): replace preceding/succeeding with front/back.
yeah, that makes sense to me.
(looking at the code also explains why "left" and "right" didn't seem as bad to the authors of this code as it does to mere users... your view of memory is quite different to ours, which only contains the actual objects we've allocated, and has no "chunks"...)
I think I'd rather keep left/right as internal concepts and only translate them to before/after in the output.
compiler-rt/lib/asan/asan_allocator.cpp | ||
---|---|---|
75 | I don't get this. Why P and S vs F and B? | |
443 | Not sure about this either. Before can be understood as in time or in space, while left and right are unambiguous, as soon as we agree on conceptual memory layout. | |
compiler-rt/lib/asan/asan_thread.cpp | ||
393 | Every time I read front/back I have to stop and think which is which. The terms make sense for a container, but IMHO not for a chunk of memory. |
PTAL. I'd like to merge this earlier rather than later if possible, to avoid merge conflicts.
compiler-rt/lib/asan/asan_allocator.cpp | ||
---|---|---|
75 | if you were referring to the below, that was a mistake (it used to be preceding / succeeding but i think front / back is better. happy to revert back to that). |
I don't get this. Why P and S vs F and B?