在Go中，为什么“ Split”不是“ string”类型的成员函数？ [重复]

Question

This question already has an answer here:

• Why are string functions not defined on the string type? 2 answers

Here are some snippets when you want to split a string with some specific delimiter in different languages:

# python
s = 'a,b,c,d,e'
tokens = s.split(',')

// javascript
let s = 'a,b,c,d,e'
let tokens = s.split(',')

// go
s := "a,b,c,d,e"
tokens := strings.Split(s, ",")

As you can see, "split" is a member function of type string in Python and Javascript, but not in Go. I am wondering why is that, it seems like STL in CPP, why the functions to manipulate an instance of a type are not member functions of that type, it seems easy to implement them in Go, like:

// go
func (s *string) Split(d string) []string {
  // here goes the code to split s with d given
}

what is the reason it is designed this way?

</div>

Answer 1

As you can see, "split" is a member function of type string in python and javascript, but not in golang.

That seems to have been so from the very beginning: commit 729bc5c, Sept 2008, for Go1 is the first commit with any mention of a string Split() function.

rudimentary string utilities.

Those functions were considered "utilities", and not part of the predeclared string type 'string' itself.

It was documented soon after in commit 0f7306b, March 2009, still Go1

// Split returns the array representing the substrings of s separated by string sep. Adjacent
// occurrences of sep produce empty substrings.  If sep is empty, it is the same as Explode.
func Split(s, sep string) []string {

You can see it used for the first time in commit 5eae3b2, April 2009 in func LookPath(file string) (string, *os.Error) {

The same approach was use for byte with bytes: commit 7893322, June 2009; Go1, with a similar Split() function.

add a bytes package analogous to the strings package.

The general idea is: you can change that utility function without changing the value type itself.
See commit 30533d6, June 2009:

Change strings.Split, bytes.Split to take a maximum substring count argument.

func Split(s, sep []byte, n int) [][]byte

An even more drastic evolution:commit ebb1566, June 2011

strings.Split: make the default to split all.
Change the signature of Split to have no count, assuming a full split, and rename the existing Split with a count to SplitN.

The other idea is to keep using string, while possibly removing dependencies to those utility functions when you don't need them (as in commit 35ace1d, Nov. 2009: "remove dependencies on strconv and strings")

It also allows to add more related function, without touching string itself.
See commit 5d436b9, Nov. 2009: lines := strings.SplitAfter(text, " ", 0), which uses Split().

Another advantage: you can optimize those functions independently of string itself, allowing for duplicate 'Split' function to be replaced by strings.Split().
See commit f388119, March 2013, Go 1.1

go/printer: use strings.Split instead of specialized code

With the faster strings package, the difference between the specialized code and strings.Split is in the noise:

benchmark         old ns/op    new ns/op    delta
BenchmarkPrint     16724291     16686729   -0.22%

The opposite case is also true: replacing strings.Split by a simpler code, as in commit d0c9b40, Sept. 2015, Go 1.6

mime: Remove an allocation in word decoding.

This fixes a TODO in (*WordDecoder).Decode by replacing a call to strings.Split with simple prefix/suffix checking and some custom slicing.

Benchmark results:

benchmark                    old ns/op     new ns/op     delta
BenchmarkQEncodeWord-8       740           693           -6.35%
BenchmarkQDecodeWord-8       1291          727           -43.69%
BenchmarkQDecodeHeader-8     1194          767           -35.76%

(same idea in commit ecff943, Sept. 2017, Go 1.11)