[wp-trac] [WordPress Trac] #57814: Use `wp_theme_has_theme_json` in `_wp_theme_json_webfonts_handler`

[WordPress Trac]

#57814: Use `wp_theme_has_theme_json` in `_wp_theme_json_webfonts_handler`
---------------------------+------------------------------
 Reporter:  spacedmonkey   |       Owner:  (none)
     Type:  defect (bug)   |      Status:  new
 Priority:  normal         |   Milestone:  Awaiting Review
Component:  Script Loader  |     Version:  6.0
 Severity:  normal         |  Resolution:
 Keywords:  has-patch      |     Focuses:  performance
---------------------------+------------------------------

Comment (by flixos90):

 @spacedmonkey @hellofromTonya I have conducted a performance benchmark for
 https://github.com/WordPress/wordpress-develop/pull/4185 in
 [https://docs.google.com/spreadsheets/d/1Oe-6ZO_Jqt6qSp2yF-
 ohb05McEB2y22aqU0S7gXuyGs/edit this spreadsheet]. I did two comparisons,
 with 250 runs each. For the first one I manually calculated the total load
 time from the two other metrics, which however is not entirely correct
 statistically, so then I also ran a second check where each of the runs
 had its own total load time.

 Summary of the results:
 * The fix here speeds up the time WordPress takes to bootstrap, before it
 starts template output **between 11-16% at the median**. The faster
 bootstrap time is expected, since by adding the extra condition here the
 logic in `_wp_theme_json_webfonts_handler()` is not run, which is hooked
 into `plugins_loaded`, i.e. runs very early.
 * At the same time, the fix slows down the time WordPress takes to render
 the template **between 9-16% at the median**. This needs more
 investigation. I was first thinking that it might make sense because of
 the `theme.json` parsing happening later now, but on second thought I
 wonder why that is happening ''additionally''. There may be something else
 that is normally calculated in here and now due to bailing in the logic
 still happens later.
 * The total load time sees an **improvement of ~1%** at the moment based
 on the statistically correct `wp-total` median.

 It is safe to say that this very small change can have wide performance
 implications, so I think we should continue to explore it. The fact that
 this is a private function that hopefully gets removed in 6.3 doesn't mean
 we cannot enhance it, particularly if there is a real performance impact
 to be made.

 That said, right now the partial regression for the template rendering
 time means we need to figure out what is happening there and why. If we
 can do that and solve the remaining problem, this ticket has the potential
 to be a significant performance enhancement.

-- 
Ticket URL: <https://core.trac.wordpress.org/ticket/57814#comment:5>
WordPress Trac <https://core.trac.wordpress.org/>
WordPress publishing platform