Feels like an especially severe version of the consistent Google pattern of only testing stuff on Chrome, so new updates/features ship in a way that is some degree of broken on Firefox/Safari. For a significant amount of time YouTube had bad performance on Firefox because they chose to use Web Components by default with a horrible polyfill instead of using the old (still working!) html5 version that ran great.
Putting in place an infrastructure to test this kind on changes on the 5-10 most popular browser would be, I think, very cheap for a company like Google. I can't help thinking these may be deliberate moves to eat the little market shares of Chrome concurrents.
I remember reading here on HN an article written by an ex-Mozilla insider relating the dissonance between the "friendly" Mozilla-Google employees exchanges and the year-long track record of very oddly recurrent "unfortunate mistakes" from Google degrading the Firefox compatibility.
Yes - they clearly don’t test the GCP console in Firefox since they “accidentally” break it on a regular basis, and there’s just no excuse for that happening at such a rich, well-staffed company.
> Putting in place an infrastructure to test this kind on changes on the 5-10 most popular browser would be, I think, very cheap for a company like Google.
The problem wasn't some web server, it's the Firefox backend services running on GCP.
Testing wouldn't have revealed anything because this didn't break with Firefox outright, it only broke when Firefox telemetry used it due to a complex series of circumstances.
