[00:21:23] which datastore backend does wikimedia use for grafana ? [00:23:05] looks like its graphite [00:23:12] Yes [13:25:20] catfuneral: graphite and prometheus mostly [14:00:28] Technical Advice IRC meeting starting in 60 minutes in channel #wikimedia-tech, hosts: @CFisch_WMDE & @bd808 - all questions welcome, more infos: https://www.mediawiki.org/wiki/Technical_Advice_IRC_Meeting [14:02:27] * bd808 has been warned :) [14:17:01] Anyone know this week's meeting topic? [14:17:49] sario528: We don't have an official one yet. I think Fisch and I will just try to help with whatever people bring up in the meeting. :) [14:18:03] Ok [14:21:13] Random suggestion: Meeting hosts should be voiced in channel during the meeting, to help identify them [14:28:36] meeting starts at what time? [14:29:35] half an hour I believe [14:30:03] yes, 15:00 UTC [14:31:16] wm-bot: next ;) [14:45:39] I want to fuss about the file-import tools [14:50:23] Technical Advice IRC meeting starting in 10 minutes in channel #wikimedia-tech, hosts: @CFisch_WMDE & @bd808 - all questions welcome, more infos: https://www.mediawiki.org/wiki/Technical_Advice_IRC_Meeting [15:02:18] \o/ [15:02:34] Welcome to the Technical Advice IRC Meeting everybody! [15:03:27] I need things I can do while waiting for jenkins [15:03:40] run dumps profiling jobs :-P [15:03:44] We have no pre-posted topics on the page so feel free and fire away with your stuff! [15:04:00] :-D [15:04:05] fire? [15:04:07] can we fire jenkins? [15:04:11] he's a slacker [15:04:19] under-resourced [15:04:53] * bd808 waves [15:05:16] I have a serious q but it can wait til others have asked [15:05:23] * apergos looks at Dragonfly6-7 for example [15:07:18] * CFisch_WMDE only knows Firefly E01-14 [15:08:05] I have questions about Wikidata [15:08:52] https://en.wikipedia.org/wiki/List_of_paintings_by_Paul_C%C3%A9zanne [15:08:53] Reedy: :p fire Jenkins and you're gonna be cooking your own pasta with butter [15:09:13] I'd like to generate this with a WD query, including the references [15:09:21] nothing wrong with home-cooked pasta with butter. now I'm hungry, thanks a lot [15:09:40] I got the table, but I don't know how to get the ref. [15:10:10] * yannf likes some pasta and butter ;oP [15:10:20] yannf: Hmmm good question. - Any Wikidata pro in here? Lucas_WMDE for example? ;-) [15:10:26] I can help with the SPARQL part for references, but I have no idea about how to integrate that with Listeria [15:10:53] so I wanted to fuss about the two tools, FileImporter and URL2Commons, both of which have a feature which would be very useful for the other [15:10:54] https://commons.wikimedia.org/wiki/User_talk:Yann/List_of_paintings_by_Paul_C%C3%A9zanne [15:10:59] for the table ^ [15:11:54] yannf: which of the statements has the interesting references? [15:12:22] Dragonfly6-7: So what's the idea? [15:12:37] let me think how to phrase this [15:12:43] P973 [15:14:10] both URL2Commons and FileImporter give the option to change the filename in the process of importing. Both tools also give you the option to then (delete the original / tag it for deletion). However, only URL2Commons automatically points to the new name. [15:14:22] this would be useful in FileImporter [15:14:24] I'd like to have [P973 Bla Bla] [15:14:30] oh [15:14:39] Also [15:14:48] so not references of statements (in the Wikidata sense), but using a main statement as a reference (in the Wikipedia / cite sense) [15:14:55] I don’t think I can help with that, sorry [15:15:35] Lucas_WMDE, yes [15:16:34] Dragonfly6-7: We're currently working on adding a process that allows automatic addition of things to the source file [15:16:35] FileImporter actually previews what the metadata will look like. URL2Commons makes sure that all the necessary fields are filled in, and that there's a proper file extension, and that the filename isn't in use, but it doesn't give you a preview of the page, and thus any mistakes - a missing { , for instance - don't get noticed until after the file has been loaded. [15:16:55] Where are the prefilled reasons pages stored? [15:17:33] Dragonfly6-7: Specifically we want to add the now-commons template / equivalent that would than also include the target's file name [15:17:36] so sometimes I'll load a dozen files in a series, and use the same set of metadata, and they'll all have the same missing { which breaks all the templates in each of them, and I'll only notice when the bot comes in and tags them all as "no license". [15:17:47] is there a way to get a WD value in Wikipedia? Then I can create a template [15:18:00] Lucas_WMDE, ^ [15:18:15] CFisch_WMDE - so you're saying that my first request is a known issue which is already being addressed? [15:18:18] yannf: https://www.wikidata.org/wiki/Wikidata:How_to_use_data_on_Wikimedia_projects ? [15:18:25] Dragonfly6-7: Yes :-) [15:18:33] CFisch_WMDE - splendid. And my second request? [15:20:04] For the second request: We're currently not in the situation to maintain / expand the URL2comons tool - but if you could add a task on Phabricator for your request and it might be picked up then by a volunteer. [15:20:42] ( if you're not familiar with Phabricator I could also create that task ) [15:20:45] Lucas_WMDE, ok, thanks. I will have a problem when there are multiple values, but I can start with that [15:22:10] CFisch_WMDE - that would be very kind of you, as Phabricator is not within my skillset. [15:22:23] CFisch_WMDE - have I phrased my request clearly enough? [15:22:25] Dragonfly6-7: Just out of curiosity: What's your use case when you upload multiple files with URL2commons? [15:22:41] CFisch_WMDE - I don't entirely understand what you're asking, sorry. [15:22:44] Dragonfly6-7: Yes, I got the idea :-) [15:22:54] 'use case' ? [15:23:02] how often does it happen, under what circumstances do I do it? [15:23:11] Under what circumstances :-) [15:23:26] Sorry for my tech slang ;-D [15:23:39] let's say I find someone's website with an archive of newspaper cartoons by a cartoonist who died in 1929. [15:23:50] "use case" is the fancy way of asking "what do you do and why?" [15:23:53] and I want to put them all on Commons [15:24:15] they all have the same creator, and the same license [15:24:47] so it's quicker to just leave those fields filled in [15:24:57] (this is not a hypothetical, i should say) [15:25:24] Ah, yeah I get it. Good example, thanks. [15:25:47] and as I leave those fields filled in, any typos - if I hold the 'shift' key down for a fraction of a second briefer than I thought, and a } becomes a ] - remain [15:25:57] and I end up with a ton of images that all get tagged as 'no license' [15:30:00] I'd like to ask about profiling, since no one else is speaking up [15:30:07] I see that URL2Commons is solo-maintained by Magnus. That makes me think that feature requests that don't come from someone willing to provide code might take a while. [15:30:40] apergos: what sorts of things are you thinking about profiling? [15:30:46] https://www.mediawiki.org/wiki/Manual:Profiling [15:30:58] oh I have done it. just that I'm sure I'm doing it 'the wrong way' [15:31:17] :) so the senior engineer mindset [15:31:19] that manual page is out of date (1.32: StartProfiler is no longer used), so I found that, ok [15:31:36] bd808 - would it be possible to emulate the feature from FileImporter within URL2Commons? [15:31:54] one is just suppose to set the global right in LocalSettings (so I did that). I'm profiling *maintenance scripts* [15:32:07] apergos: the first person I would ask about profiling is Krinkle and the second would be AaronSchulz [15:32:12] "simply reuse the relevant code", he suggested blithely, with no knowledge of how feasible it is. [15:32:59] which is not really documented (or the docs are broken/obsolete). I got whines about a bad # of args in includes/libs/Xhprof.php [15:33:39] Dragonfly6-7: FileImporter (if I'm thinking of the right thing) is part of MediaWiki. URL2Commons is a tool that Magnus wrote. There is probably no simple way for them to "share" code. [15:33:53] so I had to edit the arg list in enable() from [ $flags, $options ] to just [ $flags ] [15:34:08] and then actually add a stanza to the 'disable' function to write out the data someplace [15:34:14] Lucas_WMDE, ok, it works, but I have a problem with {{Painting table row}} [15:34:27] surely this means I'm going about this a brute force 'yeah it works but that's not how to really do it' way... [15:34:47] ah I should say: tideways-xhprof <-- that's what I'm using, apparently that's what is preferred [15:35:19] yeah, I think tideways is the only maintained fork of xhprof these days [15:35:22] bd808 - I'm thinking of the "Export to Wikimedia Commons" link [15:35:28] What would be a good task/project for someone with minimal coding experience? [15:35:40] Dragonfly6-7: The tool was written and is maintained by https://en.wikipedia.org/wiki/User:Magnus_Manske . He does his coding on some different platform than Phabricator but I will leave him a task with your suggestion there. [15:35:55] Dragonfly6-7: Can you give me your on wiki name so I can add it to the task? [15:36:03] sario528: https://phabricator.wikimedia.org/project/view/169/ might be a good place to start looking [15:36:30] DragonflySixtyseven [15:36:41] bd808: thanks [15:37:31] (which is too long a name for Freenode, and when I had my RfA, someone poached "Dragonfly67" on here, so I decided to add a hyphen. And now I'm used to it.) [15:38:54] Dragonfly6-7: Dragonfly67 is actually free now [15:39:08] yes, I know [15:39:19] as I said, I'm used to this one now. [15:39:23] Lucas_WMDE, what's the difference between {{#statements: and {{#invoke: ? [15:39:27] ah [15:41:01] Dragonfly6-7: That's the task I created for Magnus on his project https://bitbucket.org/magnusmanske/url2commons/issues/9/allow-users-to-see-a-preview-of-their [15:41:29] If he's working on it you'll probably see movement there ( due to comments added etc ) [15:41:40] that'll do, thanks [15:42:58] yannf: {{#statements:}} is a parser function from Wikibase that you can use for some simple data access [15:43:11] {{#invoke:}} is a parser function from Scribunto that you can use to invoke a Lua module [15:43:20] which may use Wikibase’s Lua functions to get data [15:43:26] but also completely unrelated things [15:43:41] ok thanks [15:50:49] oh, one other thing [15:51:09] when you import a file from enwiki to commons, the tool automatically detects the file extension and does not allow you to change it [15:51:10] but [15:51:36] when you rename a file on enwiki, the tool *does not* automatically detect the file extension, and if you mistype you can accidentally change it, which leads to an error message [15:53:08] Sooo you mean if you give the file a different extension on enwiki than what would be appropriate for the file and then try to import it? ( e.g. renaming a .gif file to '.jpg' ) [15:55:51] Lucas_WMDE, do you know where this =CONCATENATE( is used? [15:56:15] We just recently applied a check in the import, that would be triggered, if the file you try to import has a "wrong" file extension on the source wiki ( in the example enwiki ) [15:56:37] You cannot import that file unless you change the extension on the source in that case. [15:56:54] CFisch_WMDE - not necessarily importing it, just changing the filename on enwiki (for nonfree files with inappropriate names) [15:57:15] say, a historic photo of a dead person whose name has been misspelled [15:59:36] Dragonfly6-7: Hmm so your issue is generally about changing the file name on a wiki? [15:59:39] yes [15:59:59] If you want to fix the file name on the wiki and then by accident change the extension. [16:00:01] yes [16:00:10] then you get an error message and you have to do it again [16:00:14] Ok I got it know :-) [16:00:17] -k [16:01:06] Yeah might be an idea to not allow touching the file extension when renaming files on a wiki. [16:01:28] But I also could imagine that in some cases you want to change the extension :-) [16:01:44] I'll look on Phabricator and see if I can find a related request. [16:02:02] Otherwise I could add one for you. [16:03:13] when would you want to change the extension? [16:03:46] .ogg -> .ogv or .oga [16:03:58] ah [16:04:33] yannf: I don’t understand what you mean by that, sorry [16:04:47] https://commons.wikimedia.org/wiki/User_talk:Yann/List_of_paintings_by_Paul_C%C3%A9zanne [16:04:58] Lucas_WMDE, here at the bottom ^ [16:05:50] this was done by Jarekt [16:07:20] that doesn’t look like SPARQL to me [16:07:33] and I don’t think it’s Lua either [16:08:04] Lucas_WMDE: It's Excel/Spreadsheet :P [16:10:11] ok, so download as CSV, and then use a speadsheet software [16:17:15] Dragonfly6-7: Here's your task https://phabricator.wikimedia.org/T225113 ;-) [16:18:32] The official part of the meeting is over since a few minutes now. So thanks to everyone involved and see you next time! o/ [16:19:23] And as usual: Remember https://discourse-mediawiki.wmflabs.org/ for 24/7 technical advice :-D [16:28:40] I'll see if Krinkle is around later to check in with about the non-hackish way to do profiling [16:28:41] thanks a lot [16:29:05] CFisch_WMDE, very kind, thank you [17:35:14] CFisch_NA: I guess I was mostly around for moral support today. Thanks for doing most of the answering :) [18:41:33] hello! [18:42:55] i'm wondering if cumin can take into account hierarchical facts from puppet 5... [18:43:02] for example here i see: [18:43:03] # facter os.distro.codename [18:43:03] buster [18:43:11] but i can't use F:os.distro.codename=buster [18:43:57] anarcat: hey, from the top of my head no, they are not (yet) supported because it's a different API endpoint on the puppetdb side [18:44:18] right [18:44:20] okay, thanks [18:44:32] turns out F:lsbdistcodename=buster still works even though it's not listed in the facter output [18:44:35] it should be the fact-paths one IIRC [18:45:24] right, interesting one [18:46:17] yeah I can confirm that still works, as we too have recently upgraded the clients [18:47:51] I guess we'll end up adding support for it at some point, but it seems not super urgent [18:50:08] right