ResolverImplementer has a break change that the implementor can't know a field status. about gqlgen

[StevenACoffman]

@roneli I'm not quite clear on this problem. Can you please take a look?

[roneli]

Hi,

I am implemented #2850, I didn't break the original interface as I didn't want it to cause issues with backwards compatability. Basically, I concluded that not passing if it exists doesn't allow the implementer to decide / modify in future runs.

@StevenACoffman I can modify the interface to also include the prevDecl (will be empty if it was never declared etc') so the resolver implementor can decide if to change it or not. Question is are we okay with breaking the interface? I can make a new one that also accepts previous and switch case based on the provided implementation to keep backwards compatibility.

I'll dive deeper into the code and upload a proposed interface tomorrow.

[roneli]

Hi, @tsingsun

I created a PR that gives the resolver implementor the previous method body, it will be empty if no previous body was created before.

It does break the interface, but I would assume its a minor inconvenience for the anyone who uses this interface to update.

@StevenACoffman would love for any comments on the interface change