Why delete_metadata and update_metadata slow performance?

[cjj25]

I’ve been doing some investigating on the overall performance of a local WordPress with WooCommerce build and something doesn’t seem to be making sense, I’m not sure if I’m missing something or if it’s an oversight in the development.

WooCommerce has ways of checking if meta keys and values need to be updated, for example update_post_meta in class-wc-order-data-store-cpt.php

The plugin also has a method update_or_delete_post_meta for handling the meta of posts integration with the WordPress Core.

Here’s where the confusion comes in,

1. [delete_metadata](​https://github.com/WordPress/WordPress/blob/21cf92796123e98bcc9cc7981d80472977673fab/wp-includes/meta.php#L335)

Why does delete_metadata by default query to see if it exists before running the DELETE operation? I understand the DELETE operation is expensive but in this instance, we know the keys that are going to be removed. I understand we could hook into the delete_{$meta_type}_metadata
filter, return false and run the query ourselves but this doesn’t future proof things. I propose an additional filter and/or argument that allows you to bypass the initial lookup.

2. [update_metadata](​https://github.com/WordPress/WordPress/blob/21cf92796123e98bcc9cc7981d80472977673fab/wp-includes/meta.php#L159)

There is a similar problem here too. [Every update queries the database](​https://github.com/WordPress/WordPress/blob/21cf92796123e98bcc9cc7981d80472977673fab/wp-includes/meta.php#L220) to see if a row already exists and if so, then use the add_metadata. If we know the meta key 100% exists, why go to the expense of then querying the database again? Once again, I appreciate this is added as a fail-safe to ensure the data gets stored but this leaves the developer with no “official” way of updating the meta without having to do a lookup first. Again, we can hook into update_{$meta_type}_metadata but this has the same problem as before.. breaking future changes. I propose the same as above for this method too.

Has anybody else experienced this or found any solutions to excessive SQL lookups (other than relying on cache)?