While developing any go applications you might come into a situation where you need to use a module that is not yet ready to publish. The module might be living on your local machine and you wanted to get the taste of it using with other projects before shipping it. In this post, we will discuss what approach you can take to import a module that is still in development and before publishing it to the package registry.

We will write a simple project which takes a string and returns a boolean whether the string is valid IP address or not. The functionality is not that important but the important part is how we integrate it into another project while still making changes to the project.

Projects Layout

To keep things simple, I have created two projects which live side by side. The wip-module project is the unpublished one and the tester-module is the one in which will import the unpublished wip-module project. You can find the folder structure on Github.

WIP Project

The wip-module is a simple go module that holds the implementation of checking whether the IP is valid or not. Let’s look into the file:


package ipchecker

import (

func IsValidIP(host string) bool {
    if ip := net.ParseIP(host); ip != nil {
        return true

    return false

This module does not have the main package which means it was intended to be used as a library in other projects.

Tester Project

This is the project which will import the ipchecker package. Here is how it looks:


package main

import (

    ipchecker ""

func main() {
    fmt.Println("valid IP = ", ipchecker.IsValidIP(""))
    fmt.Println("valid IP = ", ipchecker.IsValidIP("invalid-ip"))

// Output:
// valid IP =  true
// valid IP =  false

As you can see it’s using the ipchecker package from the module. At this point, if you run the tester-module program you will see an error related to the package. Let’s glue them together!

Using Unpublished Modules

Before we do anything let’s look at the go.mod file of the tester-module.



go 1.16

require v0.0.0-unpublished

Since the module is not published we gave it a version (v0.0.0-unpublished) that does not exist yet but to make things work correctly.

The next step is to replace the import behaviour for the wipmodule. We can do this by telling the go compiler to replace the import for with the local path of the module. Since they are siblings, I can use relative path dot (..) notation.

go mod edit[email protected]=../wip-module
go get -d[email protected]

This will add a replacement in the go.mod of the tester-module.



go 1.16

require v0.0.0-unpublished

replace v0.0.0-unpublished => ../wip-module

Can you notice a replace directive at the end of the file? It adds the logic to tell go get to use the implementation which is one level up in the wip-module.

Now if you will run the program in the tester-module you will see this output:

$ cd tester-module
$ go run .

# Output
valid IP =  true
valid IP =  false

Note: Make sure to remove the replace directive from go.mod of the tester module before you commit your changes, otherwise go get will break in non-local environment.


We have used the golang’s module replace directive to use the local go module which is not a published module to import it into the other projects.