Upgrade to smarty 5

README.md

@@ -1,8 +1,6 @@
-# An up-to-date Smarty V4.3.5 versus Twig V3.8.0  benchmark [19 February 2024]
+# An Smarty versus Twig benchmark
 
-An *opinionated* comparison benchmark of the latest versions of Smarty v4 and Twig v3.
-
-TL;DR: Smarty is slower than Twig, Twig is faster than Smarty. Don't trust benchmarks from years ago.
+An *opinionated* comparison benchmark of the latest versions of Smarty v5 and Twig v3.
 
 ## Opinionated How?
 
@@ -13,14 +11,6 @@ The main differences from other benchmarks you might run into:
     XSS trouble.
   * Doing a benchmark that _disables auto-escaping_ can hide performance problems in a part of the template engine you
     *definitely* want to be using.
-* This test is being performed in 2017, which means we're not even going to bother with PHP5
-  * A *lot* of things are different in PHP now, for example, objects have better performance characteristics in PHP 5.4+ (https://gist.github.com/nikic/5015323)
-
-Most other benchmarks I've seen date from around 2011 or 2012. That's ancient in web-development terms. And unfortunately,
-I've seen a few PHP coders get caught out, choose Smarty because "it's faster" (or so says one benchmark), and end up without
-escaping throughout their templates.
-
-Yes, authors should probably know to escape their output. But, equally, a single oversight shouldn't result in a vulnerability.
 
 ## Templates
 
@@ -35,19 +25,23 @@ The test templates are pretty simple:
 Don't take my word for it:
 
 * `composer install`
-* `php bench.php smarty` or `php bench.php twig` or `php bench.php twig_reuse`
+* `php bench.php smarty`, `php bench.php smarty_reuse`, `php bench.php twig` or `php bench.php twig_reuse`
+
+Or, if you want to change the number of iterations (defaults to 100.000):
+
+* `php bench.php smarty 42`, `php bench.php smarty 999999`
 
 ## Results
 
-With Smarty 4.3.5 and Twig 3.8.0, on PHP 8.3, 100000 iterations, compile time ignored, cache warmed, my machine, best time
-of dozens of runs each:
+With Smarty 5.4.0 and Twig 3.11.0, on PHP 8.3, 1.000.000 iterations, compile time ignored, cache warmed, my machine:
 
-Benchmark | Time Taken
---- | ---
-twig | 1.0 seconds
-twig_reuse | 0.8 seconds
-smarty | 1.7 seconds
+| Benchmark    | Time Taken  |
+|--------------|-------------|
+| twig         | 9.2 seconds |
+| twig_reuse   | 8.5 seconds |
+| smarty       | 9.5 seconds |
+| smarty_reuse | 8.8 seconds |
 
-See the code for the difference between the twig and twig_reuse scenarios (basically: using the same Twig_TemplateWrapper instance, versus
-loading the template again. I'm not sure there's an equivalent for Smarty, but I'd be happy to include one if there is.)
+See the code for the difference between the normal and reuse scenarios (basically: using the same Template instance, versus
+loading the template again.)
 

bench.php

@@ -3,7 +3,7 @@
 
 require __DIR__ . '/vendor/autoload.php';
 
-function benchmark(string $type, int $iterations = 100000)
+function benchmark(string $type, int $iterations)
 {
     echo 'Benchmarking ' . $type . "\n";
 
@@ -37,25 +37,34 @@ function setup($type)
 {
     switch ($type) {
         case 'smarty':
-            $smarty = new Smarty();
-            $smarty->escape_html = true;
-            $smarty->compile_check = false;
+            $smarty = new Smarty\Smarty();
+            $smarty->setEscapeHtml(true);
+            $smarty->setCompileCheck(false);
             $smarty->setCacheDir(__DIR__ . '/cache');
             $smarty->setCompileDir(__DIR__ . '/cache');
             return $smarty;
+
+        case 'smarty_reuse':
+            $smarty = new Smarty\Smarty();
+            $smarty->setEscapeHtml(true);
+            $smarty->setCompileCheck(false);
+            $smarty->setCacheDir(__DIR__ . '/cache');
+            $smarty->setCompileDir(__DIR__ . '/cache');
+            return $smarty->createTemplate('index.html.smarty');
+
         case 'twig':
-			$loader = new \Twig\Loader\FilesystemLoader('templates');
-			
-			return new \Twig\Environment($loader, [
-				'cache' => __DIR__ . '/cache',
-			]);
+            $loader = new \Twig\Loader\FilesystemLoader('templates');
+
+            return new \Twig\Environment($loader, [
+                'cache' => __DIR__ . '/cache',
+            ]);
 
         case 'twig_reuse':
-			$loader = new \Twig\Loader\FilesystemLoader('templates');
-			
-			$env = new \Twig\Environment($loader, [
-				'cache' => __DIR__ . '/cache',
-			]);
+            $loader = new \Twig\Loader\FilesystemLoader('templates');
+
+            $env = new \Twig\Environment($loader, [
+                'cache' => __DIR__ . '/cache',
+            ]);
 
             return $env->load('index.html.twig');
         default:
@@ -67,9 +76,14 @@ function benchmarkOnce($type, $instance, $data)
 {
     switch ($type) {
         case 'smarty':
-            /** @var Smarty $instance */
+            /** @var Smarty\Smarty $instance */
             $instance->assign($data);
             return $instance->fetch('index.html.smarty');
+        case 'smarty_reuse':
+            /** @var Smarty\Template $instance */
+            $instance->assign($data);
+            return $instance->fetch();
+
         case 'twig':
             /** @var Twig_Environment $instance */
             $template = $instance->load('index.html.twig');
@@ -87,4 +101,5 @@ function benchmarkOnce($type, $instance, $data)
 exec('mkdir cache');
 
 $type = $argv[1] ?? null;
-benchmark($type);
+$iterations = (int) ($argv[2] ?? 100000);
+benchmark($type, $iterations);

composer.json

@@ -3,8 +3,8 @@
     "description": "Benchmark of Smarty versus Twig",
     "type": "project",
     "require": {
-        "smarty/smarty": "^4.3",
-        "twig/twig": "^3.8"
+        "smarty/smarty": "^5.4",
+        "twig/twig": "^3.11"
     },
     "license": "MIT",
     "authors": [