A backing array is the array that stores the elements referenced by a Go slice. Different slices can reference the same backing array, so changing an element through one slice can change what another slice observes. section 5.4.3
Variadic arguments can share storage
A variadic function such as testRun(args ...string) receives a local slice. Calling testRun("-c=2", "-n=1", "https://github.com/inancgumus") gives that local slice a new backing array. Calling testRun(args...) shares the input slice’s backing array. If testRun assigns args[0] = "MODIFIED", the original slice changes. Passing testRun(append([]string{}, args...)...) copies the elements into a slice with a new backing array, keeping the original unchanged. section 5.4.3
Listing 5.6 Test runner helper (hit/cmd/hit_test.go)
func testRun(args ...string) (*testEnv, error) { #1
var tenv testEnv #2
err := run(&env{ #3
args: append([]string{"hit"}, args...), #4
stdout: &tenv.stdout, #5
stderr: &tenv.stderr, #5
dryRun: true, #6
})
return &tenv, err #7
}
Inspect the test runner helper to see append create the command-argument slice while the variadic parameter receives its elements. section 5.4.3
Backing arrays also affect memory use. io.ReadAll initially reads into a newly allocated 512-byte block, then repeatedly uses append; depending on the response size, this can allocate new backing arrays multiple times. Reading an entire response therefore stores all bytes and may increase garbage-collector and CPU pressure. section 7.3.3
