In principle I agree, provided that the module path was specified explicitly.

However, if the module path was inferred automatically (for example, from the location within GOPATH), we should probably still log something about the inferred path.

CC @matloob @jayconrod

(#35070 (comment) describes another situation where the logging is probably helpful, but the proper fix there is to stop writing the go.mod file in that case.)

go mod init do print the inferred path. The problem is that it print it on stderr and with insignificant information like go: creating new go.mod:.

It may log more useful information in case it is converting from an old versioning tool.
If this information is important and should be available without a new -v flag, it could be written on stderr.

However, as I wrote, the module path, alone, should always be printed on stdout, either when the path is inferred or explicitly set by the user.

Just to clarify, the reason go mod init should print the module path of the newly created module on stdout is to allow a tool to invoke go mod init and get the module path, for later use.

I'm writing a go-init tool, that initializes a git repository at the user specified path, cd to it and invoke go mod init. It then needs the module path later, and I was forced to invoke go list -m, too.

What's wrong with invoking go list -m to determine the module path?

Nothing. But I think go mod init should do it, since I'm creating a new module.