Make type error fatal to prevent assert failure at later point #8313
Conversation
Marenz
force-pushed
the
checkFunctionsExistInIsoltest
branch
2 times, most recently
from
7a70fa6 to
5d53d3c
Compare
| @@ -2,9 +2,8 @@ contract C { | |||
| uint256 Test; | |||
|
|
|||
| function f() public { | |||
| emit Test(); | |||
| emit f(); | |||
There was a problem hiding this comment.
This is not what the test is supposed to test...
There was a problem hiding this comment.
I thought the idea was to test the code that produces the error message "must emit event" and that one is only output if you use f(), if you use Test() you only get "Test is not callable"
There was a problem hiding this comment.
Ah well, ok maybe it is, but then you can remove the uint256 Test; above.
There was a problem hiding this comment.
Or maybe best: change Test to be something callable.
Marenz
force-pushed
the
checkFunctionsExistInIsoltest
branch
from
5d53d3c to
8911b58
Compare
|
But maybe (independently of this PR) we should have a general discussion about whether to generally always make type errors fatal, whenever you cannot reasonably assign types to all expressions and document that somewhere... Especially, since in the long term and considering stuff like error recovery we might want to deal with all these cases more gracefully... |
|
Alternatively we could adopt something like an "ErrorType" used in places where the code was supposed to assign a type but errored (and similar dummy constructs) |
|
Yeah... the problem with that is that we'd probably need special cases for the error type all over... at least that could happen and wouldn't be nice... |
|
This is missing a changelog entry. |
|
Created #8316 so that we do not forget it. |
fixes #8283, #8272