[00:52:26] <snitch>	 [[Tech]]; Erik Baas; /* Empty template variables */ new section; https://meta.wikimedia.org/w/index.php?diff=25014890&oldid=25013039&rcid=27321218
[10:15:55] <snitch>	 [[Tech]]; Matěj Suchánek; /* Empty template variables */ Reply; https://meta.wikimedia.org/w/index.php?diff=25016223&oldid=25014890&rcid=27323509
[15:12:18] <snitch>	 [[Tech]]; Legoktm; /* Empty template variables */ Reply; https://meta.wikimedia.org/w/index.php?diff=25016580&oldid=25016223&rcid=27324624
[17:30:04] <roy649>	 Could somebody who understands the low-level details of how mediawiki handles oauth take a look at https://phabricator.wikimedia.org/T336610 please?
[17:30:25] <roy649>	 It looks like if you supply oauth credentials, the server doesn't handle redirects properly.
[17:30:59] <roy649>	 but I'm getting out of my depth on the low-level HTTP details.
[17:34:15] <RhinosF1>	 roy649: can you reproduce with curl
[17:34:37] <roy649>	 you mean on the command line?
[17:34:52] <RhinosF1>	 https://stackoverflow.com/questions/3044315/how-to-set-the-authorization-header-using-curl
[17:34:55] <RhinosF1>	 roy649: yes
[17:35:37] <roy649>	 No, because I don't know how to do oauth with curl, but your last message looks like something I should read :-)
[17:36:31] <RhinosF1>	 Yes :)
[17:36:46] <RhinosF1>	 That would work out if it's python or mediawiki
[17:37:15] <roy649>	 that's oauth-2
[17:45:17] <RhinosF1>	 Search OAuth1?
[17:45:30] <roy649>	 yeah, not finding much.
[17:45:36] <roy649>	 anything, really
[17:46:20] <roy649>	 my understanding is that oauth1 and oauth2 just describe different flows, but ultimately would end up with the same token.
[17:46:32] <roy649>	 so as far as curl is concerned, it shouldn't matter.
[17:46:43] <RhinosF1>	 Try it then
[17:46:46] <roy649>	 but I can't figure out how to generate a token.
[17:48:30] <roy649>	 It doesn't make sense to me that anything having to do with auth would affect a redirect, however.
[17:48:46] <roy649>	 so none of this really makes sense to me.
[17:50:21] <RhinosF1>	 It doesn't make sense
[17:51:32] <roy649>	 On the theory that it doesn't actually matter if the token is valid or not, I tried:
[17:51:39] <roy649>	 curl -H "AuthToken: 0000" -v https://en.wikipedia.org/wiki/Main%20Page
[17:51:43] <RhinosF1>	 Ye
[17:51:54] <RhinosF1>	 What response did you get
[17:52:04] <RhinosF1>	 Compared to without AuthToken
[17:52:09] <roy649>	 That gets the correct 301 to:
[17:52:10] <roy649>	 < location: https://en.wikipedia.org/wiki/Main_Page
[17:52:44] <RhinosF1>	 So Python is at fault then
[17:55:20] <roy649>	 Yeah, I guess, but the requests library is pretty rock-solid code that's been around forever and used by everything.
[17:55:32] <roy649>	 it's not the first place I would guess would be at fault.
[17:56:09] <roy649>	 maybe I'll open a bug against requests and see what they say.
[17:56:25] <RhinosF1>	 Try it
[17:56:31] <RhinosF1>	 They might say what's wrong
[17:56:46] <roy649>	 They might say I'm an idiot, but I'm used to that by now.
[17:57:02] <RhinosF1>	 That wouldn't be new for me either
[17:57:19] <RhinosF1>	 I spent about 2 days at work debugging a capital letter
[18:36:46] <roy649>	 RhinosF1 https://phabricator.wikimedia.org/T336624