GODRIVER-2589 Clarify *Cursor.All() behavior in comment.

[qingyang-hu]

GODRIVER-2589

Summary

Clarify the *Cursor.All() behavior in the comment that a nil slice pointer won't be modified in the case of no results.

Background & Motivation

The current behavior is a feature rather than a bug as long as we clarify it in the documentation.

Currently, in the case of no results:

1. if the pointer is nil, it will be left as is.
2. if the pointer points to an initialized slice, the slice will be set to empty.

The caller controls the received type either way.

Also, as mentioned in the comment, users may have depended on the behavior.

We need to merge the update to the master branch as well.

[prestonvasquez]

This behavior makes sense to me. Initializing the slice seems like strange behavior. However, it's notable that encoding/json does this:

package main

import (
	"encoding/json"
	"fmt"
	"strings"
)

func main() {
	decoder := json.NewDecoder(strings.NewReader(""))

	var data []any

	err := decoder.Decode(&data)
	if err != nil {
		panic(err)
	}

	fmt.Println(data == nil, data)
}

[prestonvasquez]

Should we also clarify the empty case?

A nil slice pointer will not be modified if the cursor has been closed, exhausted, or is empty.

[qingyang-hu]

... However, it's notable that encoding/json does this ...

The json Decode() is different from our case because returning nothing is not an exception for a query.

[prestonvasquez]

If we are checking an error can we use require.NoError? It seems there's no point in continuing the test otherwise. Additionally, we don't have to include the error in the message since the assertion will provide it:

require.NoError(t, err, "failed to create cursor")

[joyjwang]

LGTM!

[prestonvasquez]

LGTM, with suggested test updates.