[00:05:10] It's doing Parsoid... shhhhh.... don't... scare it... off.... [00:26:30] mooeypoo: at some point maybe it might be easier to install MediaWiki normally, without vagrant :) [00:27:01] legoktm: i thought about that, which is why if that was needed, i'd have reinstalled ubuntu; i've been having weird issues with LAMP [00:27:14] also, it's not that good with things I need like cross-wiki stuff [00:27:36] and then I'll need parsoid service running.... eh. It's actually not ideal [00:27:58] running parsoid is "npm install && npm start" :) [00:28:15] ok provisioning ended, there was one error with Flow but I ran composer install in Flow and then reprovisioned, all good; running update.php for all wikis anyways [00:28:25] yeah still, vagrant has very good setup if it can be used [00:29:18] OMG IT IS ALL WORKING [00:29:25] bd808: lxc fixed everything \o/ [00:29:38] I have visual editor, I have maps, I have echo and flow, and they're all working [00:29:41] and vagrant is faster than before [00:29:42] <3 [08:47:09] Hi! Has anything been changed with API:Logevents yesterday? [08:47:15] /w/api.php?action=query&format=json&list=logevents&titles=&lelimit=3 [08:47:20] seems to be new or changed -> [08:47:27] "params": { [08:47:30] "0": 176404259, [08:47:34] "1": 171268042, [08:47:38] "2": "20180412082451" [08:47:42] } [08:47:55] in "params" has been "curid" and "oldid" before, but not "0" and "1" and "2" [08:48:25] On which wiki did you see that? [08:48:26] does anybody know what happens here? [08:48:43] That'd make it easier to tell what changed there yesterday. [08:50:26] dewiki, but not enwiki [08:50:51] https://de.wikipedia.org/wiki/Spezial:ApiSandbox#action=query&format=json&list=logevents&lelimit=3 [08:51:57] eddiegp, and not frwiki too [08:52:15] seems to be dewiki only [08:52:49] * eddiegp looks [08:55:00] eddiegp: Oh, I have found it in enwiki too [08:55:17] { [08:55:18] "logid": 90221760, [08:55:19] "ns": 0, [08:55:19] "title": "Bo Burnham", [08:55:19] "pageid": 18221803, [08:55:21] "logpage": 18221803, [08:55:23] "params": { [08:55:26] "0": 836036127, [08:55:28] "1": 835994615, [08:55:31] "2": "20180412085256" [08:55:37] }, [08:55:38] "type": "review", [08:55:38] "action": "approve-a", [08:55:38] "user": "Gareth Griffith-Jones", [08:55:40] "timestamp": "2018-04-12T08:52:56Z", [08:55:42] "comment": "" [08:55:45] }, [08:56:25] seems to be "type": "review" only [08:58:30] Pastebin, please? :) [09:03:16] * eddiegp guesses FlaggedRevs [09:04:08] ... which both enwiki and dewiki have, and frwiki doesn't. [09:06:54] Hi [09:07:13] where in database are revision tags stored? [09:08:04] I need to find VE edits made in January [09:10:14] And the 'review' log action belongs to FlaggedRevs indeed. [09:12:56] i couldn't find it in the last 300 frwiki logs [09:13:56] maybe they have no "type": "review" log [09:14:20] Yeah, it's pretty sure that the deploy of "Stop logging autopatrol actions everywhere" yesterday around 13:00-14:00UTC broke something because it didn't respect FlaggedRevs [09:14:30] FlaggedRevs is the extension dewp uses for "Sichtungen" [09:14:47] frwiki doesn't have it installed [09:14:55] and enwiki? [09:15:03] has it installed [09:15:11] what is to do? [09:16:10] It provides the ability to review edits. mediawiki/core doesn't do that (or at least not to the extent FlaggedRevs allows to). [09:21:03] doctaxon: Don't know if you've seen about it in -cloud. I've pinged the one who was deploying the config change that probably did this yesterday, he'll take a look later. [10:04:26] Waggie is a fuckshit [10:04:35] Waggie wags his wagger for children [10:05:07] lol Waggie [10:05:10] !ops [10:05:14] Waggie is a faggie [10:05:27] waggin his little waggie faggie [10:08:04] someone forgot to take its medication today [10:08:18] or took the wrong meds [10:08:28] maybe be [10:08:44] he's probably drunk [10:08:48] hence the "beeeeeeeeeeeeeeer" [10:09:30] he definitively is on a beer efect (~beer@...) [10:10:09] It's nice that they alerted the channel ops of themselves with the stalkword though [10:10:39] Makes them being kicked faster :) [10:11:09] it's so good when trolls have no brains :D [10:12:38] The bot won't respond to pings from unknown users like that. [17:37:55] mooeypoo: glad to hear that LXC worked better. Debugging issues with VirtualBox is really painful. I get the feeling that it is best supported on Windows and Macs with Linux support being a bit of an afterthought. [19:18:52] bd808: I can't emphasize enough how awesome lxc is right now with vagrant. Not only did I get all the roles I wanted without trouble (VisualEditor/Parsoid were a huge problem with the other attempts) it's also *fast* [19:19:55] mooeypoo: :) it should be much faster than VirtualBox. LXC is much closer to just a bare local installation than to a virtual machine [19:21:01] Its very similar to a giant Docker container with all the parts inside of it [19:21:22] bd808: I'm trying to think how to document this properly [19:21:29] mooeypoo: tell your friends and mediawiki-vagrant skeptics! [19:21:49] yeah, its a bit tricky in the main docs [19:22:00] I'm actually very concerned; the only reason I stuck with it for almost a week was because I *really* need it for work, and creating my own server is not helpful (I need cross wiki, ORES, wikidata, things that are a nightmare to do on your own) [19:22:13] Honestly, though, if this is what volunteers need to go through, I don't see how they stick with it [19:22:23] * halfak perks up [19:22:36] the vast majority of volunteers I work with are on Windows where VirtualBox is more robust [19:22:50] but the install steps are still quite cumbersome [19:23:20] "Honestly, though, if this is what volunteers need to go through, I don't see how they stick with it" -- this applies to 90% of our processes [19:23:37] True, but I did lose a couple of interested volunteers to the dev installation issues. Mostly windows, yes, but still -- the main issue I ran into was that when they run into problems I have no idea about the Windows side, and we have very few people who can actually help [19:23:46] bd808: that too :\ [19:23:47] bd808, that applies to all of wikimedia. [19:24:00] Ever try editing :) [19:24:03] I very clearly remember saying in my first annual review that I would have left the community if I wasn't being paid [19:24:06] Yeah I guess. I'd like to see if we can get it better, though, step by step [19:24:57] there has been some real movement recently about at least acknowledging the problem [19:25:19] we don't have answers yet, but Victoria sees that this is a place we need to put in work [19:25:48] I've been tapped to at least help get discussions and planning going [19:25:59] I'll be talking at WMCON about it next weekend [19:26:17] But again, we are a long way from solutions today [19:27:09] bd808, what would it take to make a culture shift? [19:27:18] To value newcomer experiences more? [19:27:48] bd808: btw, is any team actually officially responsible for Vagrant? [19:27:52] mw vagrant [19:27:57] halfak: I think that's the same question as "how do we make enwiki more friendly to newcomers" [19:28:04] mooeypoo: not really... no. [19:28:24] I think it's not as bad as that, bd808 -- we, at least, have a code of conduct, and a smaller community that are led by people who seem to care about this shift, for the most part [19:28:32] releng was going to take it over at one point, but that never really happened [19:28:43] bd808, https://medium.com/@gmugar/preserving-the-margins-on-digital-platforms-c42bdbab8dad [19:28:53] bd808: i think that's the *first* problem. If there's no one responsible for what we consider our dev tool environment, we're relying on the good will of volunteers and... well... you [19:29:12] halfak: not to say its a bad question, just that its in the realm of chaos. We don't know what change will lead to what outcome [19:30:46] mooeypoo: its pedigree is that ori started the project because he felt that onboarding new team members in E3 was too hard. RobLa encouraged me to use it from my first day. At some point Ori handed "control" to me. From there things get muddy [19:31:32] there are a lot of folks like Matt Flaschen, tgr, and dduval that have put major work into it [19:31:50] but there is no canonical owner or paid maintainer [19:32:34] RelEng has (had?) a FY18/19 program to work on developer tools [19:32:43] but they are looking into a different solution [19:33:07] they want to work on Docker + Kubernetes tooling that can reuse things built for production deploys [19:33:34] basically a "built, test, and deploy the same code and config" solution [19:34:05] It would be sweet if that works, but until then I'll try to keep the wheels on the bus in my 'spare' time [19:34:22] CindyCicaleseWMF has some plans of trying to merge the various strains of setup / dev environment work (MW installers, docker, Vagrant, meza) next FY I believe [19:35:04] or rather, merge the requirements and then try to come up with one tool for all [19:36:24] Hello! Reading back to see what I missed . . . [19:36:51] hey CindyCicaleseWMF. mostly just some rambling about difficult developer experiences :) [19:37:51] Ah, yes, I feel your pain ;-) [19:37:55] mooeypoo found out about using MediaWiki-Vagrant with an LXC runtime yesterday which was both exciting for her and saddening because she had fought with VirtualBox for days/weeks/months [19:39:20] halfak: I've skimmed that article a couple of times now. I need to really sit down and read it to find out if it overlaps with Yuvi and Summana's talks about accidental complexity or if that's just a connection I make for all technical interventions [19:39:43] I'm working on making that connection now with J-Mo [19:39:57] The literacy angle is kind of funny. [19:40:23] I don't think the complexity being incidental matters so much [19:40:36] It's really any barrier to participation [19:41:11] right. The core of the accidental complexity argument for me is basically that existing structures and processes get in the way of the important work. [19:41:19] But the fun thing is tying it together with successors (Harding/Harraway) and boundary objects (Bowker & Star) [19:41:29] and often those things are not really needed [19:41:34] I think it also gets in the way of newcomers being able to alter alter culture. [19:42:02] s/newcomers/anyone who isn't currently an authority-powerful-person/ [19:42:40] The re-interpretation of what we're doing and how we do it is important for evolution and adaptation -- to solve the problems of today. [19:43:04] The lit around bureaucracies likes to call this "mediation" work. [19:43:29] E.g. we might all agree about WP:Verifiability, but how we apply that is mediated by how people interpret the importance of WP:V [19:43:56] Similarly, we all value security, robustness, open source. But the way we enact that can become entrenched. [19:44:23] bd808: btw, speaking of lxc, I wasn't alone in moving to it -- RoanKattouw was next to me when I watched how a page loads in my new mw-vagrant, and he got so excited about how fast it is, he immediately switched to lxc too :D [19:44:24] If we "open up the margins" we allow others to mediate those shared values differently. [19:45:05] The reason why all these fancy words matter: It provides a theoretical strategy for enabling change without taking a patriarchal approach to implementing it. [19:45:24] * bd808 pushed his POV from the top down [19:45:38] mooeypoo: nice! [19:45:40] By building up toolforge and paws, we make a margin for non-authoritative people to re-mediate what it means to build tools and do data science. [19:46:06] PAWS and Quarry are really our main attempts to enter this space I think [19:46:08] Those people have different needs and concerns that come closer to the front and center of how we make decisions. [19:46:36] ToolForge is a good example too. Who do you know who is a great part-time sysadmin (and not also a fulltime sysadmin)? [19:46:37] Toolforge is a high learning curve that really only removes a relatively small financial barrier [19:46:56] that's true. managed services are helpful [19:46:58] btw, bd808 another point -- 1. For some annoying reason, all my repos were https:// rather than ssh:// ... I switched them all, but a newbie might not know how to? I am 99% positive I ran the script, but I should try again in a separate machine to see if there's an issue with it for newbies. 2. I keep having to set 'git branch --set-upstream-to=origin/master master' in all repos. Any chance that can also be pre-prepped ? [19:47:12] * halfak goes back into paper-writing cave [19:47:13] o/ [19:47:15] but you can get them from Heroku or other places as well [19:48:42] mooeypoo: The https fetch by default is done on purpose, but there is some hidden instruction on how to make all clones with ssh. The big issue is that to do that you have to forward your ssh key into the VM and let Puppet use it which is not the world's best opsec [19:49:02] Why is that on purpose? Aren't we assuming this is a dev environment? [19:49:28] https means you don't need to set up SSH keys [19:49:41] yes, but giving random Puppet code from MediaWiki-Vagrant use of your ssh key is the scary part [19:49:47] And I didn't have to do anything other than 'git remote rm origin && git remote add origin ssh://[username]@gerrit.wikimedia.org:[whatever the port]' [19:50:05] I do 'git review' from the host [19:50:06] I guess vagrant could use HTTPS to fetch but keep the git config in SSH but it seems dirty [19:50:17] mooeypoo: have you tried running `vagrant git-update` since doing that? I bet it blows up [19:50:21] oh [19:50:24] ... no [19:50:40] or any vagrant role that installs something new [19:51:14] meh, but I'm thinking a newbie that joins in and needs to submit a patch to, say, Echo [19:51:28] you'd install mw-vagrant per instructions, you have it working, you work on the patch, and you can't submit it [19:51:31] https://www.mediawiki.org/wiki/MediaWiki-Vagrant#Pushing_commits is a nice solution for that [19:51:42] oh shoot i missed that one [19:51:43] git-review does the right thing as far as I know with an http remote [19:51:57] It didn't for me, but maybe I was missing some config [19:52:22] but yeah, the config is a good solution [19:52:32] it should certainly be advertised better [19:52:35] Maybe it should be higher up [19:52:37] yeah [19:52:58] * mooeypoo runs that line for her vagrant [19:53:17] I was chastised by an unnnamed person yesterday for running "random commands from the web" [19:53:26] heh [19:53:34] In my defense, it wasn't random, I knew what it was doing, but apparently linux treats it worse than I would guess or something [19:53:41] curl some_random_url | sudo bash [19:53:52] haa not that bad [19:54:24] it was apt-get dist-upgrade [19:54:33] which I read about, found to be reasonable, and ran it [19:54:53] but apparently that is a scary one? It should be noted in apt-get --help as "SCARY" or something [19:55:00] I have `[url "ssh://gerrit/"] insteadOf = https://gerrit.wikimedia.org/r/p/` in my ~/.gitconfig and some magic in my ~/.ssh/config to make ssh://gerrit/ use the right port, username, and key [19:55:23] well, it's an OS upgrade [19:55:37] I don't think it's scarier than other ways of upgrading your OS [19:55:44] tgr: yes, but I do "yes" to those when they pop up in the GUI [19:55:50] dist-upgrade can be unpredictable, but if you look at the plan it makes and agree then it's fine IMO [19:55:57] so... I didn't think it was that bad running those as a precursor to installing something [19:56:15] as in, the "how to install X in ubuntu" gave that as a step, I read through what it does, thought it reasonable, ran it [19:56:32] And then the next step failed, and I was told that the dist-upgrade may have blown stuff up [19:56:34] ah. yeah, that's a bad tutorial [19:56:44] Is this the code review office hours? [19:57:06] They do usually go 'apt-get autoremove' and such so I thought it was reasonable to make sure your system is updated fully before installing the thing [19:57:14] but apparently there's an extra sense to linuxing [19:57:19] HunterH: technically, not the conversation we are having, but do you have some code that you need to find a reviewer for? [19:57:37] HunterH: not sure that exists anymore, but you can try getting reviewers at any time, that's (partially) what this channel is for [19:58:30] mooeypoo: re the git branch thing, that shouldn't happen - vagrant does a git clone, that is supposed to set upstream automatically [19:58:32] bd808: I do, and per the Phabricator event (https://phabricator.wikimedia.org/E413) it says there is a #mediawiki-codereview channel held weekly to get +2 reviewers to view gerrit submissions. [19:59:05] there used to be [19:59:30] HunterH: *nod* I think that Mukunda stopped doing that meeting at some point, but please do share your patch :) [19:59:33] tgr: yeah, but it's not the first time I have to manually set it :\ [19:59:59] not sure if it's something I missed (I did do a WHOLE bunch of attempts) or if it's something missing in vagrant when it configures itself, or something like that [20:00:01] Thanks! https://gerrit.wikimedia.org/r/#/c/408577/ [20:00:03] was killed per T173770 and T177974 I think [20:00:04] T173770: Code Review Hours advertised but not taking place? - https://phabricator.wikimedia.org/T173770 [20:00:04] T177974: Drop #wikimedia-codereview channel - https://phabricator.wikimedia.org/T177974 [20:00:37] someone should cancel that event [20:00:37] multimediaviewer! fun times. [20:01:11] HunterH: "Gergő Tisza" on that patch is tgr here ;) [20:01:51] tgr? haven't seen him [20:02:01] tgr: looks like the "repeat until" on it stopped 2018-03-23 [20:02:33] I am new to HFOSS, and have been working on this issue (T161612)for a while. Hopefully will get my first merge soon :-) [20:02:34] T161612: Buttons in MMV are not really buttons and are thus not semantic - https://phabricator.wikimedia.org/T161612 [20:02:42] uh I'm trying to distance myself from mediaviewer-related things [20:03:12] rgt: fair enough. Should we troll marktraceur with it? [20:03:33] or Volker_E? [20:03:56] there is a team for that now, and the existing tasks are either low-impact or blocked on people having strong feelings about it [20:04:08] Volker_E for that specific patch, yeah [20:04:55] Volker_E: HunterH is hoping to get more feedback or a merge on https://gerrit.wikimedia.org/r/#/c/408577/ [20:05:27] make a new contributor happy if you can! :) [20:08:09] Please and thank you in advance. [20:09:09] HunterH: sorry this is stalled out for you. This particular code is outside my area of comfort for merging. [20:10:36] bd808: np. Thanks for the help.