Don’t Reuse That []byte Buffer

Go’s bytes.Buffer is an excellent tool to hold temporary data. Except, you need to remember the difference between string and []byte . Take a look at the sample below, and see if you can deduce how it will behave without running it.

The Go Playground

There two things to note:

1. Reset() resets the underlying []byte buffer in bytes.Buffer, but it reuses for subsequent reads and writes.
2. In Go, a string is immutable, and you can only create new ones, whereas slices share the same underlying storage

This means that (bytes.Buffer).Bytes() and (bytes.Buffer).String() have slightly different semantics. The return value from the former is a slice with shared storage with the the bytes.Buffer object, where as the latter returns a completely standalone object.

So in our previous example, you will see that foobytes will contain the same data that barbytes contain, except for the fact that foobytes is the view into the same []byte with len=13 (the length of “Hello, World!”) as opposed to len=15 (the length of “Goodbye, World!”)

The takeaway here is, if you need to use the result of writing to bytes.Buffer, but want to reuse bytes.Buffer object, you will need to copy the data that the underlying buffer in bytes.Buffer to another []byte or string in order to make it independent from subsequent modifications to the object.

Hope that helps someone in the future. Happy (Cautious) Go Hacking!