[Lldb-commits] [lldb] [lldb][NFC] Refactor Watchpoint class (PR #163695)

[via lldb-commits]

jimingham wrote:

You say:

`While I agree with your point that "the low level Process code says why we stopped, and sets an appropriate StopInfo", my key point is that the current implementation can't always correctly derive StopInfo. Specifically, in my example, the correct StopInfo should have been StopInfoTrace, not StopInfoWatchpoint.
`

But so far as I can tell, the only reason you knew that was the "correct StopInfo" was because the thread plan system's current plan was step over breakpoint and it was EXPECTING a stop reason of trace (or maybe because the ThreadPlanStepOver was waiting for it which would be even more tenuous)  

I think that having the thread plan stack be both responsible for determining the stop reason and reacting to it would make for a system that would get very very hard to reason about very quickly.


https://github.com/llvm/llvm-project/pull/163695