[15:48:52] multichill: ping. potter, an aspiring GSoC student, wants to ask you about the image recognition idea. I told him you don't have time to mentor, but he still wants to talk with you about it. he is in #mediawiki [19:38:29] @MediaWikiMeet [19:42:06] Why this command doesn't runs here? [19:42:14] @list [19:42:19] oh [20:02:33] DarTar: for the disable-button-click event [20:02:46] if you click the disable button while the flyover is already open, it closes it [20:03:04] do you want to track that independently? [20:12:20] hey rsterbin [20:12:20] hey [20:12:28] I only care about three events: click on the close button next to the feedback link, impressions and clicks on the gotoprefs button [20:12:42] ok [20:12:46] clicks to dismiss the dismiss flyover (ha), I don't care [20:13:03] it's the same button; that's why i ask [20:16:36] I see, we should definitely count only the clicks on the feedback link button (A or E), not the flyover [20:16:36] ok [20:16:36] thanks [20:16:36] so if i click the X once to open the flyover, then click it again to close the flyover, only one event will be registered [20:16:36] well, one click event and one impression event ;) [20:16:36] RoanKattouw or anyone else who knows about logging from MW on the cluster: one of the mobile devs has some API code that's causing a fatal on enwiki and nowhere else. he wants me to hack some debug output around the offending line and was suggesting a var_dump(); die(); kind of thing. but is there some way we can just log arbitrary output somewhere we can view? [20:18:09] AaronSchulz, do you know? ^ [20:18:18] awjr: is this using the clicktracking extension and writing to the log on emery ? [20:18:28] DarTar: nope [20:18:31] ok [20:18:44] it's related to MobileFrontend [20:18:49] awjr: you can add trigger_error( string ) to dump it to the error log [20:19:06] AaronSchulz: cool - where can we view the error log? [20:19:44] tail -f /home/wikipedia/syslog/apache.log [20:20:24] AaronSchulz: awesome thanks [20:21:00] Yeah that works [20:21:16] You can also throw an exception, which will show up in exception.log with a backtrace [20:21:50] RoanKattouw: maybe I should make a custom debug log group for stuff like this [20:22:04] for wfDebugLog() [20:22:21] Would be nice [20:22:54] The currently favored (and undocumented) hack is to use wfErrorLog( 'Message', 'udp://10.0.5.8:8420/foo' ) which logs to /h/w/logs/foo.log provided that file exists and is world-writable [20:28:15] hey ChristianWikia - do you have an interest in https://www.mediawiki.org/wiki/Summer_of_Code_2012#Mentor_signup ? I think we talked in person a couple weeks back. [20:33:55] sumanah: Got several questions about people who want to work on opencv..... [20:34:11] multichill: multiple people? [20:35:14] Two topics at https://commons.wikimedia.org/wiki/User_talk:Multichill#Tagging_images_in_OpenCV and got an email some time ago [20:37:03] multichill: can you suggest anyone who could guide them or mentor them? [20:37:54] Not really [20:38:36] multichill: then you should tell them that [20:40:28] RoanKattouw: it seems throwin an exception in this case is not working as its getting caught by the API [20:40:33] or maybe im doing something wrong [20:43:47] No, you're right [20:43:56] trigger_error() should still work right? [20:50:47] trigger_error() doesn't appear to be outputting an errors to apache.log either :( [20:50:54] RoanKattouw, AaronSchulz ^ [20:51:15] but at least we can now see the exception in the resulting json from the api call [20:51:24] is the event actually happening? [20:51:28] yes [20:51:39] did you sync the file? [20:51:50] yup [20:51:57] so this is what i added: [20:51:58] http://pastie.org/3630604 [20:52:13] sine i wasn't seeing anything in the log, i also now have it throwing an exception [20:52:27] which you can now see getting thrown here: https://en.wikipedia.org/w/api.php?action=query&prop=excerpts&format=json&exlimit=10&titles=Barack%20Obama [20:52:38] Well because of the exception, the trigger_error call will never be reached [20:52:45] oh duh. [20:52:51] :p [20:52:52] And vice versa [20:53:10] RoanKattouw: not unless he had USER_FATAL [20:53:13] oh trigger_error() will stop execution also? [20:53:20] by default trigger_error just warns and keeps going [20:53:38] Oh [20:53:40] ok so theoretically i should be able to use both if i do trigger_error(); throw new Exception() [20:53:48] Yeah, but only in that order [20:54:00] rightright [20:56:36] ok i reversed the statements and yeah still not seeing anything in /home/wikipedia/syslog/apache.log from trigger_error() but i think he's able to get the info he needs from the exception coming back in the api response [23:42:09] awjrichards@fenari:/home/wikipedia/common/php$ scap "Pushing changes to MobileFrontend per https://www.mediawiki.org/wiki/Extension:MobileFrontend/Deployments#19_March.2C_2012" [23:42:09] Checking syntax... [23:42:09] /usr/local/bin/lint: line 2: 1536 Segmentation fault php -n -dextension=parsekit.so `dirname $0`/lint.php "$@" [23:42:09] Found syntax errors in 1.19, cannot sync. [23:42:18] anybody know anything about that ^? [23:43:00] TimStarling, AaronSchulz, Reedy, RoanKattouw ^ [23:44:15] I'll look [23:44:22] thanks TimStarling [23:45:30] /home/wikipedia/common/php-1.19/extensions/WikimediaMaintenance/cleanupBug31576.php [23:45:34] it has conflict markers in it [23:45:45] not sure why that would make it segfault, but that's the problem [23:46:08] I'll just revert [23:46:27] that's a nice failure mode, if the code has an error, lint segfaults ;) [23:46:49] should work now [23:47:04] well, better than silently returning success [23:47:20] thanks TimStarling looks good