Detect silent failures (no REVERT or error message) #4
Conversation
CMajeri
force-pushed
the
feature/call_failures
branch
from
f06ee8e
to
4c6d403
Compare
| stack := scope.Stack | ||
| for i := l.depth - (depth - 1); l.depth > depth-1; l.depth, i = l.depth-1, i-1 { | ||
| if l.current.error == nil { | ||
| switch stack.Data()[len(stack.Data())-i].Bytes32()[31] { |
There was a problem hiding this comment.
It's a bit hard for me to follow the two loop iterators in my head. If I understand correctly, we always have i == l.depth - (depth - 1), right? Maybe instead of tracking i, we can calculate
stackIndex := len(stack.Data()) - (l.depth - (depth - 1))
on each iteration.
This would also allow us to make a bounds check on the variable instead of crashing out-of-bounds if for some reason the three variables scope.Stack, depth and l.depth get out of sync. What do you think?
There was a problem hiding this comment.
And what do we do when it happens?
NOT crash?
There was a problem hiding this comment.
Consider making the code clean enough such that the fatal error that is raised will allow some insight into what went wrong. I'm not aware of the context in which this is run (the filename says "logger", but this is not logging code), so if you say a fatal crash is appropriate so be it.
If you could even include a test, that would be amazing.
|
@CMajeri is this PR still relevant? Could you link the Jira that tracks the issue? |
https://taurusgroup.atlassian.net/browse/DEV-18502