You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, if we consider this as indeed a breaking change, it looks like we can detect that this is not "oneof" (which isn't clear for users) but rather the usage of field presence. So, we can change the error message to make it clear. e.g.
Field "2" on message "CheckTextRequest" became "optional" which internally moves it inside a oneof
If possible, it would be great if we could make this particular check disableable, because in our case this is absolutely fine
The text was updated successfully, but these errors were encountered:
In v1 of the buf breaking rules (which we are committed to not breaking), we took the simple approach and said that any changes around oneofs were both breaking for generated sources and wire compatibility. In reality, there is a bit more nuance around wire compatibility, which in a v2, we would look to clean up.
I agree that the error message could be better, and the breaking change rules that produce errors should be cleaned up here. I've attempted to address this in #1978
We've got the following
If we add "optional" we will get a really confusing error:
According to protobuf folks, adding "oneof" is fine (link: protocolbuffers/protobuf#10822)
However, if we consider this as indeed a breaking change, it looks like we can detect that this is not "oneof" (which isn't clear for users) but rather the usage of field presence. So, we can change the error message to make it clear. e.g.
If possible, it would be great if we could make this particular check disableable, because in our case this is absolutely fine
The text was updated successfully, but these errors were encountered: