Like the title says:
The resolver code has catches that check whether another resolver already used the given address.
Now, when switching between using the reimplemented version of a function and not, the debug build seems to sometimes fail to properly re-build the the structure, likely because of all the trickery we use.
As a result, two resolvers end up on the dll and an error is thrown. Cleaning the build directory removes the error, which makes me believe it is related to the headers tricks.
I have no idea if there is a way to enforce this better. At the moment, it has not much relevance, that might change in the future, however, when we switch to testing the remaining mismatches at least for full logic match.
For now it is a reminder.
Like the title says:
The resolver code has catches that check whether another resolver already used the given address.
Now, when switching between using the reimplemented version of a function and not, the debug build seems to sometimes fail to properly re-build the the structure, likely because of all the trickery we use.
As a result, two resolvers end up on the dll and an error is thrown. Cleaning the build directory removes the error, which makes me believe it is related to the headers tricks.
I have no idea if there is a way to enforce this better. At the moment, it has not much relevance, that might change in the future, however, when we switch to testing the remaining mismatches at least for full logic match.
For now it is a reminder.