Previously, `wpdb::get_table_from_query()` would not find the correct table name in the query `DELETE a FROM table a`, due to not recognising the table alias immediately after the `DELETE` as correct syntax.
Fixes#37660.
git-svn-id: https://develop.svn.wordpress.org/trunk@38507 602fd350-edb4-49c9-b593-d223f7449a82
This filter allows the shortcode generation process to be short-circuited, so expensive short codes can be cached and returned immediately.
Props ideag.
Fixes#37906.
git-svn-id: https://develop.svn.wordpress.org/trunk@38506 602fd350-edb4-49c9-b593-d223f7449a82
This new filter allows the smilies array to be modified with a filter, instead of having to directly access the global.
Props mte90, jorbin.
Fixes#35905.
git-svn-id: https://develop.svn.wordpress.org/trunk@38504 602fd350-edb4-49c9-b593-d223f7449a82
Links within lines or blocks of text should always be underlined since they can't
rely on color alone to be distinguished from the surrounding text. Exceptions
can be handled on a case-by-case basis.
Props Presskopp for the initial patch.
Fixes#37527.
git-svn-id: https://develop.svn.wordpress.org/trunk@38503 602fd350-edb4-49c9-b593-d223f7449a82
The third parameter of `date_i18n` is a bool. Currently, it's a weak check, otherwise the call showing the universal time on the "General Settings" screen would be messed up. Use an actual bool so we call our own functions correctly.
Fixes#37634.
Props fronaldaraujo.
git-svn-id: https://develop.svn.wordpress.org/trunk@38502 602fd350-edb4-49c9-b593-d223f7449a82
Cleans up a few places where the CSS coding standards were not being followed.
Fixes#37576.
Props Presskopp, johnpgreen, netweb
git-svn-id: https://develop.svn.wordpress.org/trunk@38501 602fd350-edb4-49c9-b593-d223f7449a82
After [38364], the inclusion of `class-wp-locale.php` was changed from a `require_once()` to a `require()`. This caused problems for anything that called `load_text_domain_early()` prior to the `require()` but didn't bail, as it was now being `require()`-ed a second time.
With the use of `require_once()`, it doesn't really matter where it's loaded, so it has been moved next to the related `l10n.php` load.
See #37827.
git-svn-id: https://develop.svn.wordpress.org/trunk@38496 602fd350-edb4-49c9-b593-d223f7449a82
Fixes issue whereby a user would see nothing happen if the pane is collapsed while they shift-click to edit an element in the preview.
Props curdin, celloexpressions, westonruter.
See #27403.
Fixes#36678.
git-svn-id: https://develop.svn.wordpress.org/trunk@38492 602fd350-edb4-49c9-b593-d223f7449a82
Makes more clear users can set either a city or a UTC timezone offset.
Props GrantDerepas, andrew.p.
Fixes#34789.
git-svn-id: https://develop.svn.wordpress.org/trunk@38483 602fd350-edb4-49c9-b593-d223f7449a82
When a customizer construct (panel, section, control) is not added in PHP, the JS has interpreted this to mean that a given construct should be deactivated (because it is gone). This is problematic for dynamically-created constructs in JS, as it has meant that the construct would also have to be created in PHP to ensure the `active` callback is called, or else a hack would be required to add a `construct.active.validate = function() { return true };` to forcibly prevent the construct from getting deactivated.
These workarounds can be eliminated by treating constructs differently when they are created dynamically in JS (after page load) as opposed to being created statically in PHP (on the server). Namely, if a construct is dynamically-created then its absence in a preview refresh should not signal that the construct should be deactivated. Rather, a dynamic construct should only have its activation state toggled if it has a corresponding construct created in PHP when the preview refreshes to explicitly indicate its `active` state. Otherwise, the management of the `active` state for a construct created in JS should also be the responsibility of client-side code.
Props westonruter, sayedwp.
Fixes#37270.
git-svn-id: https://develop.svn.wordpress.org/trunk@38464 602fd350-edb4-49c9-b593-d223f7449a82
This private query var allows the hierarchical page query on the Pages listing screen to query for `wp_posts.ID, wp_posts.post_parent` instead of `wp_posts.*`. This introduces large memory and time savings when the site contains a large number of Pages. Combined with the processing time savings introduced in [31730] this makes the Pages listing screen considerably more performant.
Fixes#34982
Props rodrigosprimo
git-svn-id: https://develop.svn.wordpress.org/trunk@38451 602fd350-edb4-49c9-b593-d223f7449a82
The SQL query was built using the clauses compiled when querying for
top-level comments. But in cases where the top-level comment query
results are already in the cache, the SQL clauses are not built, and
so are unavailable for `fill_descendants()`. Instead, we call
`get_comments()`, using modified versions of the parameters passed
to the main `WP_Comment_Query` class.
Props Akeif, Rarst for testing.
Fixes#37696.
git-svn-id: https://develop.svn.wordpress.org/trunk@38446 602fd350-edb4-49c9-b593-d223f7449a82
[37320] corrected some behaviour in how PHP and MySQL character sets are matched up. This was correct, but had the side effect of causing some incorrectly configured sites to start failing.
Prior to [37320], if `DB_CHARSET` was set to `utf8mb4`, but the PHP version didn't support `utf8mb4`, it would fall back to the default character set - usually `latin1`. After [37320], the `SET NAMES` query would force MySQL to treat the connection character set as `utf8mb4`, even if PHP wasn't able to understand it.
By checking if `mysqli_set_charset()` succeeded, we can simulate the old behaviour, while maintaining the fix in [37320].
Props danielkanchev fo helping to diagnose this issue.
Fixes#37689 for trunk.
git-svn-id: https://develop.svn.wordpress.org/trunk@38441 602fd350-edb4-49c9-b593-d223f7449a82
There were some cases where some local variables in `wpdb::init_charset()` could be undefined when they were used.
Props Frozzare.
Fixes#37683 for trunk.
git-svn-id: https://develop.svn.wordpress.org/trunk@38439 602fd350-edb4-49c9-b593-d223f7449a82
Running the taxonomy array through `array_unique()` is unnecessary
when the function returns objects, because the associative keys already
ensure uniqueness.
This also fixes a bug when running `get_attachment_taxonomies()` in
HHVM, which doesn't like casting objects to strings for the purposes
of `array_unique()`.
Props swissspidy.
See #37368.
git-svn-id: https://develop.svn.wordpress.org/trunk@38437 602fd350-edb4-49c9-b593-d223f7449a82
