I’ve been reading through all the conversation and issues here. I am interested in supporting new formats and improving performance, but I think this change being pushed by default to users when they upgrade to 6.1 is a lot for right now, including with some of the clunky interactions OSes still have around webp (and HEIC!) files.
I’m happy for support for working for webp and HEIC files to stay in core, as we should be liberal in what we accept and work with, but not with the change to convert everything to webp when JPEGs are uploaded.
During today’s Performance Team meeting, contributors briefly discussed the revert.
“We are still trying to figure out what a canonical plugin is exactly, and if that would work for WebP by default,” Google-sponsored core committer Adam Silverstein said. “We still have a couple of fixes to land for 6.1 around image quality when WebPs are output (which is still an option you just need a plugin for now).”
During the previous Performance team meeting Silverstein said Mullenweg’s post about the feature not shipping was a surprise to the team and that they were working with the release leads to better understand concerns in hopes of finding a path forward to include it in the update.
“I want to acknowledge this is a blow for everyone who worked on the feature (myself included) and at the same time I’d like to encourage us to focus on how we can move forward given the current position,” Silverstein said. “Are there concerns we can address? Does a canonical plugin make sense?”
Participants in that discussion had expressed concerns about getting users to adopt WebP by default if it is moved to its own plugin at the time. It would require strategic rebranding to indicate that it delivers faster images for the visitors, as most users won’t be familiar with the WebP format at this point.
“I think if it remains a ‘feature project,’ it makes sense to remain in the Performance Lab plugin – we don’t know if moving it out of it would get us more testers (especially since the Performance Lab plugin has 10k+ installs which is a lot for a feature plugin),” Google-sponsored contributor Felix Arntz had said.
“If the path of a ‘canonical plugin’ should be pursued, then of course we would need to take it out, but then the nature of the project would also change.”