Lets say I have an awesome Go binary that denormalizes some data, and I'd like to place it under:
//some/path/denormalize/
Ideally my package name would be denormalize which would allow me to follow the guidelines for "[when] designing a package, consider how the two parts of a qualified identifier work together, not the member name alone". Thus a potential member function might be denormalize.FromX(...). However, if I place the main.go for the denormalize binary under denormalize/ package I can't use the intended package name as now the package would be named main.
Some options I've considered:
main.go under denormalize/ and then place the rest of the code under internal/ or api/ or pkg/. Pros: This seems to be what most folks do. Cons: The package name is not meaningful, and it violates the principle of insuring that the two parts of the qualifier work together since api.FromX(...) no longer makes sense. This leads to more verbose names such as api.DenormalizeFromX(...).main.go under denormalize/main/ and the rest of the code under denormalize/. Pros: Package name stays meaningful. Cons: a subdirectory then has a dependency on a parent directory which has some code smell.Is there other options that I have not considered for where to place the main.go such that it won't force me to use a different package name?