How does Go resolve seemingly same types?-CodePudding

I have a PostgreSQL database with 2 columns: id (UUID), company_url (varchar) with following values:

2fc35af4-5f5c-445e-86c5-01d93513b8be | https://test.com
2bf31b75-d1f3-4a9c-a530-73b714816e9e | https://test2.com

Here's the code to access the table using Go (error handling and credentials omited for simplicity):

package main

import (
    "fmt"
    "database/sql"
    "github.com/google/uuid"
    _ "github.com/lib/pq"
)

func main() {
    connStr := "host= password= port= dbname= user="
    db, _ := sql.Open("postgres", connStr)
    rows, _ := db.Query("SELECT * FROM companies;")

    for rows.Next() {
        // var id [16]byte // not ok
        var id uuid.UUID // ok
        var companyURL string

        rows.Scan(&id, &companyURL)
        fmt.Println(id, companyURL)
    }
}

There also exists a Go package for UUIDs. In its' source code the UUID is defined simply as

type UUID [16]byte

In the code above the id is typed using uuid.UUID, I tried to replace the type declaration in my code above (commented out with not ok) but instead of the correct value it returns an array of 16 zeroes. uuid.UUID returns correct id.

So my question is why such behaviour if uuid.UUID and [16]byte are of the same types? There are no binaries in uuid package nor init() function, nor is Scan() making any implicit changes.

CodePudding user response：

Strictly speaking, they are not the same types. They have the same underlying type though. UUID is different in that it implements the interfaces sql.Scanner and driver.Valuer, so it can be used with the databases transparently.