coral

...but that fifth dependency makes all the difference! $ go mod init foo && go get github.com/spf13/cobra && wc -l go.sum 758 go.sum $ go mod init foo && go get github.com/muesli/coral && wc -l go.sum 10 go.sum (Number of lines in go.sum != number of transitive dependencies, of course, but the scale of the difference still hints at how much extra is being pulled into your project.)

Right but that last one is the problem... https://github.com/muesli/coral/blob/master/go.sum https://github.com/spf13/cobra/blob/master/go.sum

I like this from the readme: It's easy to use Coral as a drop-in replacement for Cobra in existing projects, just let gofmt do the heavy lifting for you: gofmt -w -r '"github.com/spf13/cobra" -> "github.com/muesli/coral"' . gofmt -w -r '"github.com/spf13/cobra/doc" -> "github.com/muesli/coral/doc"' . gofmt -w -r 'cobra -> coral' . go mod tidy Having implemented something to do a similar one off refactor in java for thousands of files, that is a breath of fresh air. In java i had to go code -> AST -> perform manipulations -> generate code from AST (with inevitable long PR’s due to generated code not matching 1:1 with parsed original). After having gone through that i think if i had to repeat again with java I’d probably try harder to use structural find and replace.

4 ... of its 5 dependencies. It's just missing Viper. We should do a better job with the title here. See: - [0] https://github.com/muesli/coral/blob/master/go.mod - [1] https://github.com/spf13/cobra/blob/master/go.mod