Performance Chat Summary: 5 December 2023

Table of Contents

Announcements

@joemcgill #58196 has a PR that could use another set of eyes, but can probably be committed. However, turns out this does not solve a big performance issue, but instead the parent function needs to be looked into instead.
@swissspidy For i18n performance I am nagging people for code reviews again at the moment. Otherwise there hasn’t really been any progress.
@joemcgill I was reviewing #55595 earlier today, and noticed that widget registration itself does not really account for much of a performance cost, but rather rendering widgets that can be costly. I think we can likely close that ticket as `maybelater` but would like a 2nd opinion.
- Additionally, I noticed that you can already use the `load_default_widgets` filter to keep WP from loading default widgets, but doing so will cause a fatal if you don’t also unhook `wp_widgets_init`, because that function will try to register the default widgets even if they haven’t been loaded. This seems like a bug, although not a critical one.
@thekt12 Removing call to build_template_part_block_variations on the fontend showed a decent performance improvement -> https://core.trac.wordpress.org/ticket/59969#comment:8 Additionally we did identify two more palaces where the same improvement could be made. Currently I am addressing PR review comments.

@westonruter The PR which adds a Partytown module to opt-in to running scripts in a Web Worker got a couple reviews.

@westonruter Pull request for applying the image optimizations is in review and nearing merge (to the feature branch)
- I also have sub-PR opened which makes breadcrumb processing more reliable. (In other words, an XPath for a given image that needs optimization.) Also ready for review.
- Now that the feature is fully functional I’m going to add test coverage (which I neglected before getting a working prototype in place).
- I also want to extend the existing logic to also handle CSS background images, whereas currently only img elements are optimized.
- I’ve also started filing issues under the module’s label for areas where others can contribute.
@joemcgill Still working on a proposal for improving the default sizes implementation in WP using available layout data. Also, seeing some progress being made on the browser side implementing the new auto sizes HTML spec, which would be a huge improvement for WP, since we lazy-load many images by default.
@swissspidy As for images, I have been looking a bit more into client-side image compression (in Gutenberg) lately to see where this could fit in 2024 plans.

@swissspidy No update either from my side, though I just published a small blog post to help spread the word a little bit about Plugin Check

@pbearne just https://core.trac.wordpress.org/ticket/42441
- @joemcgill Last I chatted with @flixos90 about this, he wanted to make sure we got feedback from some external contributors about the proposed change to the DB values. I think he’s planning to raise this in a dev-chat soon (perhaps tomorrow)?
@joemcgill Just for heads up. I saw https://core.trac.wordpress.org/ticket/60011 come across yesterday. I can see @johnbillion is reviewing, but worth keeping an eye on.
- @swissspidy Unfortunately not much information there except “there are cache misses”