[16:48:25] * AaronSchulz Krinkle: Hey! You said "You'd want to limit to the route that generates new images, not the GET for svgs" regarding math endpoint deprecation headers. What would be the harm in those emitting the header? Were you thinking of /media/math/formula/{hash} as well? [17:06:32] AaronSchulz: Only in so far that we use that endpoint heavily in prod today, embedded in articles on enwiki. If the implementation doesn't "do" anything and that consequence is fine, then go for it I guess. [17:07:08] i.e. it doesn't log a warning, or increment a counter [17:07:20] or is visible in the body or browser devtools [17:09:18] * AaronSchulz Krinkle: so, you're thinking from a flood/risk perspective. The idea would be to just set the header, which would show in devtools like any response header. [17:10:11] * AaronSchulz In any case, probably no one will check the for headers anyway. [17:16:50] Yeah, I guess if it's a custom header that we make up, it wouldn't stand out there or cause load/confusion. [17:17:42] I do note that even for third parties, a deprecation on the SVG endpoint is less actionable since that's essentially a blackbox URL returned by the formula endpoint. [17:18:22] So we wouldn't want to make them chase down a way to avoid that endpoint, they need to focus on not calling the POST endpoint and then the rest takes care of itself. [17:19:06] If they got the svg image from elsewhere, then I guess they do eventually need to chase down where that's coming from. E.g. if they are mirroring Wikipedia content or calling WMF api.php and getting it from there, then in theory that will go away by itself later this month. [17:19:23] But again in most cases it's someone else fixing it, not them. [17:19:39] anyway, no hard reason not to, something to think about :) [20:49:06] Krinkle: ok, thanks [21:40:22] https://make.wordpress.org/core/2026/09/08/join-the-phpunit-test-cleanup-for-7-2/