[19:42:07] <legoktm>	 could someone take a look at https://gerrit.wikimedia.org/r/#/c/337539/ ?
[20:03:38] <anomie>	 legoktm: So the plan there is to not bother with recursive updates on the assumption that whoever is running the script will run it on those other pages too if necessary?
[20:03:54] <legoktm>	 yes
[20:04:17] <MaxSem>	 +2'd
[20:04:29] <legoktm>	 thanks
[20:05:07] <anomie>	 oh, MaxSem beat me to it.
[20:10:31] <RainbowSprinkles>	 Heh, so the headers-already-sent we've been trying to track down? I think I found a proximate cause. Namely... we start responding to a request, have sent headers (and maybe partial content), but then we hit an exception and swap to sending an error page. The exception handler sends its own headers :)
[20:10:59] <RainbowSprinkles>	 The actual stacktrace I saw pointed to Redis being the "cause" of the failure, but the underlying workflow seems like it could be pretty easily triggered.
[20:12:41] <RainbowSprinkles>	 Hmm, it supposedly checks for headers_sent()
[20:12:45] <RainbowSprinkles>	 Curious
[22:16:14] <James_F>	 Krinkle: https://kangax.github.io/compat-table/es5/