您可以使用DB::listen() 测试您的应用程序通过listening to the queries 进行的查询。
我已设置以下内容作为测试:
迁移:
Schema::create('posts', function (Blueprint $table) {
$table->bigIncrements('id');
$table->string('title');
$table->string('content');
$table->timestamps();
});
Schema::create('comments', function (Blueprint $table) {
$table->bigIncrements('id');
$table->unsignedBigInteger('post_id');
$table->string('text');
$table->timestamps();
});
型号:
class Post extends Model
{
protected $guarded = [];
public function comments()
{
return $this->hasMany(Comment::class);
}
}
class Comment extends Model
{
protected $guarded = [];
public function post()
{
return $this->belongsTo(Post::class);
}
}
测试:
$post = Post::create([
'title' => 'Hello World',
'content' => 'Here I am!',
]);
$queries = collect();
DB::listen(function ($query) use ($queries) {
$queries->push($query->time);
});
for ($i = 1; $i <= 3000; $i += 1) {
Comment::create([
'post_id' => $post->id,
'text' => 'Facade '.$i,
]);
$post->comments()->create([
'text' => 'Relation '.$i,
]);
}
$totalTime = [0, 0];
foreach ($queries as $idx => $time) {
$totalTime[$idx%2] += $time;
}
return [
'facade' => $totalTime[0],
'relation' => $totalTime[1],
];
这个输出:
array:2 [▼
"facade" => 1861.3
"relation" => 1919.9
]
所以你可以看到create的关联方式在我的测试场景中实际上慢了大约3%。
如果你想进一步试验,我已经准备好了这个implode。