[wp-trac] [WordPress Trac] #48453: Regression: Implied contract between image sub-size filenames and their base filename now broken

[WordPress Trac]

#48453: Regression: Implied contract between image sub-size filenames and their
base filename now broken
-------------------------------------+---------------------
 Reporter:  ianmjones                |       Owner:  (none)
     Type:  defect (bug)             |      Status:  new
 Priority:  normal                   |   Milestone:  5.3
Component:  Media                    |     Version:  trunk
 Severity:  major                    |  Resolution:
 Keywords:  needs-patch 2nd-opinion  |     Focuses:
-------------------------------------+---------------------

Comment (by Iulia Cazan):

 Replying to [comment:2 slackbot]:
 > ''This ticket was mentioned in [https://make.wordpress.org/chat/ Slack]
 in #core by azaozz.
 [https://wordpress.slack.com/archives/core/p1572303558428200 View the
 logs].''

 From my perspective, moving the "original" would only complicate things,
 and that is because currently (in 5.3-RC2 also) there is no information
 recorded in the database of it, no file path, no width/height, just
 nothing, and we rely on the directory mentioned in the metadata for the
 full size to identify it (or at least I could not find another way to do
 this). It will just become a nightmare to delete it or just use it to
 generate files as a failback when something went wrong with the full size.

 From all my tests, the issue with the full-size naming happens only when
 the threshold kicks in, and the sub-sizes are not generated from the new
 full size generated, but from the original itself.

 For example, this is quite odd for big files, let's say the uploaded file
 is `image.jpeg` of 6MB, this is then generating with the new threshold
 filter a full-size called `image-2560.jpeg` of 1MB (if the script does not
 fail). The sub-sizes are then generated from the 6MB file, processing
 which is again prone to script fails not to mention it will take a lot of
 time.

 Why aren't the sub-sizes generated from the new full-size which is smaller
 and also the quality high enough as far as I could test?

 Also, what is the point to keep the "original" when the threshold was
 applied? As far as I understand the process `image-2560.jpeg` can be
 easily renamed back after the real original file and the real original
 removed (this is never used as far as I can see in the code). If the code
 would to this switch, then I do not see a problem anymore for the naming
 of the sub-sizes, and also, the sub-sizes will generate faster, from a
 smaller file.

 Did I get this right?

-- 
Ticket URL: <https://core.trac.wordpress.org/ticket/48453#comment:3>
WordPress Trac <https://core.trac.wordpress.org/>
WordPress publishing platform