[11.x] Blade Component Loop Speed Improvement

[lonnylot]

Cache repeated processes that will not change creates a 7% performance improvement:

Before:

OPCache Enabled: Enabled
With caching: 206.043693

After:

OPCache Enabled: Enabled
With caching: 192.26193209091

Profiling methodology: #51141 (comment)

welcome.blade.php

@for($i = 0; $i < 1000; $i++)
    <x-avatar :name="'Taylor'" class="mt-4" />
@endfor

avatar.blade.php

<div>
    Hi! My name is {{ $name }}
</div>

routes/web.php

Route::get('/bench', function () {
    $a = array_map(
        fn () => Benchmark::measure(fn() => view('welcome')->render()),
        range(0, 10)
    );

    return 'OPCache Enabled: ' . (is_array(opcache_get_status()) ? 'Enabled' : 'Disabled') .
            '<br>With caching: ' . (array_sum($a) / count($a));
});

[Jubeki]

This syntax should also be possible:

Suggested change

	if (isset($this->normalizedNames[$name])) {
	return $this->normalizedNames[$name];
	}
	$this->normalizedNames[$name] = ViewName::normalize($name);
	return $this->normalizedNames[$name];
	return $this->normalizedNames[$name] ??= ViewName::normalize($name);

[taylorotwell]

Hi @lonnylot - could you fix the conflicts here?

Update: Think I got it.

[lonnylot]

Hi @lonnylot - could you fix the conflicts here?

Update: Think I got it.

Pushed the fix before I saw the update

[lonnylot]

👀 Slight mistake in the updated formatting that affects performance. Let me know if you want me to correct it or if you're taking care of it (don't want to cause conflicts)

[taylorotwell]

@lonnylot you can fix.

[lonnylot]

@taylorotwell Actually your update is better and I just adjusted the tests. I can come back later and do additional profiling to confirm the performance improvement still exists.

[taylorotwell]

@lonnylot in my testing locally the performance improvements are still in place. 👍

For reference, rendering 20,000 components locally takes ~98ms before your change and ~90ms after. Taking average of 10 iterations per benchmark.