[00:06:48] I guess I can move them in master [00:14:50] Krinkle: https://gerrit.wikimedia.org/r/682779 [00:19:43] thx [04:40:09] Krinkle: a few comments on https://gerrit.wikimedia.org/r/c/mediawiki/core/+/677730 too [06:05:42] phedenskog: 🚀 [06:06:15] the update is rolling out now, will be interesting to see it appear in RUM data [06:06:51] all in all it took a year since Noam started working on it [06:30:08] the amount of FCP recordings for Safari have already increased in the last couple of days. I imagine they roll out the update early to some [09:14:56] Thinking maybe its time to move from firstPaint to FCP since we have more browsers that support that now. [10:28:34] I wrote a patch to expose Paint Timings in Prometheus this morning. For synthetic testing, I agree, we should switch [10:28:45] and track LCP as well [10:28:58] for Google reasons [12:46:27] I updated the dashboards so for Chrome on WebPageReplay and WebPageTest we have the metrics there. For alerts I haven't switched, but switching Speed Index to LCP would be good for Chrome I guess. I can create the an issue for that. [15:33:18] Krinkle: how feasible would it be for the navtiming extension's JS to be the *only* JS module running for grade C browsers? to collect before/after metrics for the impact of grade C downgrade [15:34:48] maybe that behaviour can be a sort of grade of its own (almost C?), used only during a transition period for browsers about to be phased out [15:35:23] gilles: not feasible. but even if it was, we'd need to either make all its dependencies and mw.loader/base/jquery es3 compliant and use post-ie6 features. but more feasible, which I guess is effectively the same, is to copy part of it out of the module and make it not a module but a regular script appended to inline. [15:35:42] use no* post-ie6 web features. [15:36:20] we can use a skin hook for that, and we do that for a few rare use cases already. incl the NORLQ.push() layer which emulates