[12:02:57] phedenskog: when the real device lab is up and running, a nice mini project we could do on it would be an assessment of AVIF impact on perf if all images on an article were AVIF [12:03:43] since we're seeing yet another article https://jakearchibald.com/2020/avif-has-landed/ that fails to dig into that matter, I think there's an opportunity to show some real numbers on real hardware and latest version of browsers [13:24:26] ah cool [14:05:59] gilles: once it has even decent adoption and support in Thumbor, it might make sense to use that as the only alternative and not support webp even at the same time, since the difference is so big, and webp isn't always. [14:06:18] was reading this article the other day which makes an interesting case not to use webp always, https://mastodon.technology/@krinkle/104447814913768778 [14:06:25] https://siipo.la/blog/is-webp-really-better-than-jpeg [14:06:49] does our pipeline use mozjpeg for jpegs? [16:59:31] no, I recall that the most extreme jpeg compressors took a very long time to run. not workable for on-demand rendering like we do [16:59:41] same reason we don't optimise our PNGs as much as we could [16:59:59] almost all of those article ignore both compression time and decoding time [17:00:12] file size isn't everything [17:53:31] Ack, right. Until/asynchronous, not really worthy exploring [17:53:41] until/if * [17:59:29] pretty much. the same might be true for AVIF as well, we have to check [18:00:05] webp was deliberately not as efficient as it could be at the time because of the encoding/decoding time tradeoff [18:44:58] https://www.mediawiki.org/wiki/Security/SOP/Access_to_Phabricator_Security_Issues [19:49:35] greg-g: when should the "Quarterly skip level" mtg be? [19:51:57] AaronSchulz: quarterly :P But seriously, I can reschedule that one for next week or so [19:55:02] next week works (now that vacation is over)