Version 3 of the AWS SDK for PHP

Last October, we announced the Developer Preview of Version 3 of the AWS SDK for PHP. We even presented about it at AWS re:Invent last November. We are grateful for your early feedback and support. Since last fall, we’ve been hard at work on improving, testing, and documenting Version 3 to get it ready for a stable release. We’re excited to announce that Version 3 of the AWS SDK for PHP is now generally available via Composer and on GitHub.

Version 3 of the SDK (V3) represents a significant effort to improve the capabilities of the SDK, incorporate over two years of customer feedback, upgrade our dependencies, improve performance, and adopt the latest PHP standards.

What we’re excited about

We’ve made many improvements to V3, even since our blog post about the Developer Preview (check out that post if you haven’t already). There are also some things that have changed or have been removed since Version 2 of the SDK (V2). We encourage you to take a look at our V3 Migration Guide for all the details about what has changed.

V3 has less code and better performance than V2 and is using the latest version of the Guzzle HTTP library. It also has some exciting new features and improvements.

Asynchronous requests and promises

V3 allows you to execute operations asynchronously. This not only means that it is easier to do concurrent requests, it’s also easier to create asynchronous and cooperative workflows. We use Promises, the basic building block of our asynchronous features, all throughout the SDK’s core. We also use them to implement the higher-level abstractions SDK, including Command Pools, Paginators, Waiters, and service-specific features like the S3 MultipartUploader. That means that almost every feature of the SDK can be used in an asynchronous way.

To execute an operation asynchronously, you simply add “Async” as a suffix to your method call.

// The SYNCHRONOUS (normal) way:

// Executing an operation returns a Result object.
$result = $s3Client->putObject([
    'Bucket' => 'your-bucket',
    'Key'    => 'docs/file.pdf',
    'Body'   => fopen('/path/to/file.pdf', 'r'),
]);

// You can access the result data from the Result object.
echo $result['ObjectURL'];

// The ASYNCHRONOUS way:

// Executing an operation asynchronously returns a Promise object.
$promise = $s3Client->putObjectAsync([
    'Bucket' => 'your-bucket',
    'Key'    => 'docs/file.pdf',
    'Body'   => fopen('/path/to/file.pdf', 'r'),
]);

// Wait for the operation to complete to get the Result object.
$result = $promise->wait();

// Then you can access the result data like normal.
echo $result['ObjectURL'];

The true power of using asynchronous requests is being able to create asynchronous workflows. For example, if you wanted to create a DynamoDB table, wait until it is ACTIVE (using Waiters), and then write some data to it, you can use the then() method of the Promise object to chain those actions together.

$client->createTableAsync([
    'TableName' => $table,
    // Other params...
])->then(function () use ($client, $table) {
    return $client->getWaiter('TableExists', [
        'TableName' => $table,
    ])->promise();
})->then(function () use ($client, $table) {
    return $client->putItemAsync([
        'TableName' => $table,
        'Item' => [
            // Item attributes...
        ]
    ]);
})->wait();

Please take a look at our detailed guide on promises for more information.

PSR-7 compliance and decoupling of the HTTP layer

The PHP-FIG has recently announced the acceptance of PSR-7, a “PHP Standard Recommendation” that defines interfaces for HTTP messages (e.g., Request and Response objects). We have adopted these interfaces for how we represent HTTP requests within the SDK, and it has allowed us to decouple the SDK from Guzzle such that V3 will work with both Guzzle 5 and Guzzle 6. It’s also possible to write your own HTTP handler for the SDK that does not use Guzzle.

The SDK defaults to using Guzzle 6 to perform HTTP requests. Guzzle 6 comes with a number of improvements, including support for asynchronous requests, PSR-7 compliance, and swappable HTTP adapters (including a PHP stream wrapper implementation that can be used on systems where cURL is not available).

JMESPath querying of results and paginators

In V3, the Result object has a new method: search(). With this method you can query data in Result objects using JMESPath expressions. JMESPath is a query language for JSON, or, in our case, PHP arrays.

$result = $ec2Client->describeInstances();
print_r($result->search('Reservations[].Instances[].InstanceId'));

JMESPath expressions can also be applied to Paginators in the same way. This will return a new Iterator that yields the result of the expression on every page of data.

$results = $s3->getPaginator('ListObjects', [
    'Bucket' => 'my-bucket',
]);
foreach ($results->search('Contents[].Key') as $key) {
    echo $key . "n";
}

Time to code

We hope you will enjoy using Version 3 of the AWS SDK for PHP. Here are the links you need to get started: