Continuously check code coverage while developing tests in go

I have this workflow where while developing tests I also check for code coverage. The idea is to see if I am really touching that piece of code I am adding tests for. Similar to TDD, there’s a certain primal endorphine rush from seeing the coverage go from red to green.

In a nutshell, it works like this:

1

go test ./... -coverprofile=coverage.out && go tool cover -html=coverage.out

The first bit will create a coverage.out file, which we will then process using go tool cover and generate a HTML.

It also calls, on Linux, xdg-open which opens up the file in your (supposedly) preferred browser.

However, by default it creates a temporary HTML file (on Linux, under /tmp). For example, file:///tmp/cover617771716/coverage.html#file0. So every time you run the command again, it will generate another file, which will open in another tab, making the experience quite annoying.

A simple solution is to, instead of letting cover generate a random file, to output to a fixed location file, like coverage.html, which you can then refresh manually in your browser.

The whole thing then becomes:

1
2
3
4

go test ./... -coverprofile=coverage.out && go tool cover -html=coverage.out -o coverage.html

# you only have to run this once
xdg-open coverage.html

Theoretically one can add a server that watches that file and refreshes the html.

I may do that sometime.