Async indexing #98
Async indexing #98
Conversation
# Conflicts: # lib/class-sp-post.php # lib/class-sp-sync-manager.php
# Conflicts: # lib/class-sp-post.php # lib/class-sp-sync-manager.php
This reverts commit 088ddb3.
|
phpcs errors are being fixed in this PR #99 Once that is merged in we can update this PR |
|
This is affected, and at least somewhat blocked, by WordPress/gutenberg#12903. If this PR is merged as-is, there will potentially be a very small window for a race condition where,
There are ways we could work around this, e.g. by setting the value on the sync flag to |
# Conflicts: # .travis.yml # lib/class-sp-post.php # lib/class-sp-sync-manager.php # tests/test-api.php # tests/test-faceting.php # tests/test-general.php # tests/test-indexing.php # tests/test-integration.php # tests/test-mapping-postmeta.php # tests/test-mapping.php # tests/test-searching.php
| } | ||
| $to_delete[] = absint( $post_id ); | ||
|
|
||
| update_option( 'sp_delete', array_unique( $to_delete ), 'no' ); |
There was a problem hiding this comment.
Potential overwrite race condition here. Need to figure out a better way to store this data if we're to proceed with this route. Maybe deletes should be synchronous, though?
|
Added in |
| $sync_meta = SP_Sync_Meta(); | ||
|
|
||
| // phpcs:disable WordPress.DB.DirectDatabaseQuery | ||
| $post_ids = $wpdb->get_col( "SELECT SQL_CALC_FOUND_ROWS `post_id` FROM {$wpdb->postmeta} WHERE `meta_key`='_sp_index' LIMIT 500" ); // WPCS: cache ok. |
There was a problem hiding this comment.
This should probably be DISTINCT post_id in retrospect.
Async Indexing: Remove call to delete post if unindexable because race condition
# Conflicts: # lib/class-sp-sync-manager.php # lib/class-sp-sync-meta.php
This PR replaces synchronous indexing with async indexing. It is a breaking change.
Closes #39, Refs #38, #42, #43
Update 2/2/2024
This is a stable branch in good shape. It was paused due to the Gutenberg issue. I lost touch with that issue, but it's been a long time and perhaps there's a workaround at this point.
I have another note about using
DISTINCT post_idthat should be verified and applied.Otherwise, this should get a fresh set of eyes and let's ship it as isolated as possible in its own release (ideally 0.6).