mirror of
https://github.com/gosticks/wordpress-develop.git
synced 2025-10-16 12:05:38 +00:00
b6fde03ef7
This changeset adds support for the `fetchpriority` attribute, which is typically added to a single image in each HTML response with a value of "high". This enhances load time performance (also Largest Contentful Paint, or LCP) by telling the browser to prioritize this image for downloading even before the layout of the page has been computed. In lab tests, this has shown to improve LCP performance by ~10% on average. Specifically, `fetchpriority="high"` is added to the first image that satisfies all of the following conditions: * The image is not lazy-loaded, i.e. does not have `loading="lazy"`. * The image does not already have a (conflicting) `fetchpriority` attribute. * The size of of the image (i.e. width * height) is greater than 50,000 squarepixels. While these heuristics are based on several field analyses, there will always be room for optimization. Sites can customize the squarepixel threshold using a new filter `wp_min_priority_img_pixels` which should return an integer for the value. Since the logic for adding `fetchpriority="high"` is heavily intertwined with the logic for adding `loading="lazy"`, yet the features should work decoupled from each other, the majority of code changes in this changeset is refactoring of the existing lazy-loading logic to be reusable. For this purpose, a new function `wp_get_loading_optimization_attributes()` has been introduced which returns an associative array of performance-relevant attributes for a given HTML element. This function replaces `wp_get_loading_attr_default()`, which has been deprecated. As another result of that change, a new function `wp_img_tag_add_loading_optimization_attrs()` replaces the more specific `wp_img_tag_add_loading_attr()`, which has been deprecated as well. See https://make.wordpress.org/core/2023/05/02/proposal-for-enhancing-lcp-image-performance-with-fetchpriority/ for the original proposal and additional context. Props thekt12, joemcgill, spacedmonkey, mukesh27, costdev, 10upsimon. Fixes #58235. git-svn-id: https://develop.svn.wordpress.org/trunk@56037 602fd350-edb4-49c9-b593-d223f7449a82 |
||
|---|---|---|
| .. | ||
| data | ||
| includes | ||
| tests | ||
| multisite.xml | ||
| README.txt | ||
| wp-mail-real-test.php | ||
The short version:
1. Create a clean MySQL database and user. DO NOT USE AN EXISTING DATABASE or you will lose data, guaranteed.
2. Copy wp-tests-config-sample.php to wp-tests-config.php, edit it and include your database name/user/password.
3. $ svn up
4. Run the tests from the "trunk" directory:
To execute a particular test:
$ phpunit tests/phpunit/tests/test_case.php
To execute all tests:
$ phpunit
Notes:
Test cases live in the 'tests' subdirectory. All files in that directory will be included by default. Extend the WP_UnitTestCase class to ensure your test is run.
phpunit will initialize and install a (more or less) complete running copy of WordPress each time it is run. This makes it possible to run functional interface and module tests against a fully working database and codebase, as opposed to pure unit tests with mock objects and stubs. Pure unit tests may be used also, of course.
Changes to the test database will be rolled back as tests are finished, to ensure a clean start next time the tests are run.
phpunit is intended to run at the command line, not via a web server.