[14:58:26] [telegram] We just launched [[WM:Techblog]], a new place to share stories from the Wikimedia technical community with the world. Yay!🎉 [14:58:26] [telegram] https://techblog.wikimedia.org/ [14:58:27] [telegram] https://lists.wikimedia.org/pipermail/wikitech-l/2020-March/093250.html [15:01:17] [telegram] I thought Phabricator already included one :) [15:01:50] [telegram] (At least I've been reading posts of interest there for a while) [15:05:30] [telegram] yeah, I also don’t understand how this relates to https://phabricator.wikimedia.org/phame/ [15:28:47] [telegram] the Parsoid post seems to be a copy of Phame one [15:42:12] [telegram] Yes! We found that people were posting in many different places and wanted to provide a more central and visible venue for people from the Wikimedia technical community to share about their work. [15:42:13] [telegram] We will be cross-posting some Phame posts to the new blog, as there are many intersections in audiences, and the Parsoid/PHP post is an example of this. [15:51:05] [telegram] Mandatory xkcd reference: https://xkcd.com/927/ [16:45:10] Maarten: point taken, *but* we have andre_ on the Tech Engagement team so when we make an N+1 thing he pokes every week to find out when the N-1 thing will be shutdown. We haven't talked in depth about the future of Phame yet, but I think it is reasonable that we do soon. :) [16:50:43] [telegram] I might be wrong but I think when Phabricator was introduced the idea was that it would aggregate all technical things in it. Unfortunately the progress is bad, for instance we still have Gerrit which for me and for other people who are sort of used to GitHub/Gitlab/Bitbucket not to mention to people unfamiliar with git altogether is a very unfriendly place in technical sense, but I think it is not a [16:52:44] If you think gerrit is bad for submitting code... You don't want to see what phab does... [16:54:44] Though, I don't ever remember phab being an aggregator for all tech things... Yes there was talk to move code review etc to it, but that got declined a while ago [16:54:56] Also, the git/svn argument... Well, that sailed nearly a decade ago now [17:00:01] [telegram] oh. well. I do not know indeed. but gerrit is definitely bad. from the need to use some weird git plug in to, what I have discovered only last week, that one must have no more than one commit per patch. (re @wmtelegram_bot: [irc] If you think gerrit is bad for submitting code... You don't want to see what phab does...) [17:00:11] [telegram] not to mention the web ui. [17:00:16] You don't have to use a weird git plugin [17:00:29] [telegram] What weird git plugin? [17:00:39] [telegram] git review I think? [17:00:47] you can use `git push origin HEAD:refs/for/master` [17:01:01] [telegram] I don't think that's ever been required [17:01:47] The commit per patch thing is a valid complaint [17:01:50] But it's a different workflow model [17:01:52] Doesn't make it wrong [17:01:58] [telegram] is it not needed anymore? it's been a while since I've committed to gerrit from terminal, and that was back when I did not have an okay experience with git from other sides of my life [17:02:10] As per Alex, it was never "needed" [17:02:15] It just can make things easier [17:02:18] [telegram] oh. interesting. it was in a tutorial I was using (re @Alex: I don't think that's ever been required) [17:02:49] Gerrit is not GitHub. No doubt about that. Gerrit has some problems, but honestly so does GitHub and all the other DVCS hosting services that try to clone it. I had high hopes for Differential at one point, but I used it enough to prove to myself that is was actually worse than Gerrit for ease of use (and that was a low bar to crawly under) [17:03:08] Yeah, it's "recommended", but dosn't mean it's needed [17:04:18] using `git-review` is recommended because it hides the magic branch naming for submitting a patch, but it is totally possible to use gerrit with "pure" git [17:05:20] [telegram] https://www.mediawiki.org/wiki/Gerrit/Tutorial now that I reread it it indeed mentions that git review is just recommended, so you are right. I would still make it more prominent probably in some way [17:05:31] But honestly complaints about that are strange to me as GitHub is far more difficult to automate Pull Request creation from [17:06:12] Making tutorials document every alternate workflow is not reasonable [17:06:51] That just leads to walls of text that newcomers have difficultly finding the substance in [17:07:21] [telegram] Well, I totally get why some people would like to squash a PR/MR on its merge, but keeping just one commit during the work is definitely not something that makes things comfortable (re @wmtelegram_bot: [irc] But it's a different workflow model) [17:07:22] I know this is the wiki-way in some sense, but its bad practice [17:08:13] [telegram] Well, as one does something basic it might actually be better to have less automation (re @wmtelegram_bot: [irc] But honestly complaints about that are strange to me as GitHub is far more difficult to automate Pull Request creation from) [17:08:14] Gerrit is a patch based workflow rather than a feature based workflow. Both are valid ways of organizing work in a DVCS [17:09:43] [telegram] Back when I was first interacting with Gerrit I was purely WIndows user and executing every next terminal command was something that could discourage me from doing anything. While it is definitely pretty easy to go to web UI and press a few buttons there. (re @Thecladis: Well, as one does something basic it might actually be better to have less automation) [17:10:44] Thecladis: I won't argue that Gerrit is easy to learn or obvious. It is not. [17:11:05] git on windows is hardly pleasureable at the best of times [17:11:13] And github is possibly only better with their fancy gui client [17:11:25] tortoisegit was never as good as tortoisesvn [17:11:47] GitHub has invested a lot in their web UI too for "simple" code changes [17:12:12] Simple being relative, hence the scare quotes [17:12:19] Simple changes in gerrit are pretty "easy" these days too via the web interface [17:12:26] if you know where you need to start to get the "create change" button [17:13:03] I can see paladox arguing that one, but I would love to see how many hours it takes a new user to find the button ;) [17:13:26] * bd808 sends paladox <3 for his work on Gerrit upstream [17:13:43] :) [17:13:49] [telegram] well, that is true, that was a big relief for me when I learned that it supports some that. (re @wmtelegram_bot: [irc] Simple changes in gerrit are pretty "easy" these days too via the web interface) [17:14:06] It has supported it for quite a while now though... I think [17:14:08] (and soon file uploads will be supported in gerrit :)) [17:14:11] I lose track. Years get blurry [17:14:17] [telegram] Gitlab also has a decent WebIDE fwiw (re @wmtelegram_bot: [irc] GitHub has invested a lot in their web UI too for "simple" code changes) [17:14:36] bd808 PG now has a guide for new users. [17:14:50] (its basic, and only shows if you have no changes) [17:15:25] [telegram] I think it also uses some confusing terminology there IIRC (re @wmtelegram_bot: [irc] I can see paladox arguing that one, but I would love to see how many hours it takes a new user to find the button ;)) [17:19:34] [telegram] oh yeah, branch is actually the branch you create your own off, and the actual branch you create is a topic [17:21:47] [telegram] (I was caught in that confusion at lease twice) [17:25:39] [telegram] https://www.mediawiki.org/wiki/Phabricator/Plan well so indeed, that is true, a blog space was not included in the plan, but CR and Bug reporting was among a few other things that surround the two. Still, for me personally it makes more sense to keep all stuff together :P [17:28:25] [telegram] I'm not sure Phame was ever something deliberately installed [17:28:32] well thats one point, another point that could be made is if phabricator goes down, users can still commit changes to gerrit same for the opposite. Also code review in differential is severely less powerful then gerrit. [17:28:41] [telegram] It was just part of Phab, some people found it and started publishing on it? [17:31:22] @ Thecladis also alot of work has been done on the new UI, so it may seem gerrit hasn't changed much, thats only because the UI isn't the default yet :) [17:32:34] [telegram] you mean it is possible to switch to some newer one? [17:33:04] yup [17:33:10] ?polygerrit=1 [17:33:46] though (the UI has changed quite visually since 2.15): see https://gerrit-review.googlesource.com/c/gerrit/+/258653 [17:34:21] [telegram] Btw Gerrit does not support 2FA does it? [17:34:40] [telegram] Think we have a task somewhere about that [17:34:55] It doesn't, no [17:35:26] https://phabricator.wikimedia.org/T198813 [17:35:51] [telegram] Oh right I made it [17:36:29] lols [17:36:37] [telegram] looks fresher indeed (re @wmtelegram_bot: [irc] ?polygerrit=1) [17:36:57] [telegram] is it just a different theme or an actual new UI? [17:36:58] [telegram] There is that SSO system ops were working on... Wonder if Gerrit could be integrated with that [17:37:04] an actual new UI [17:37:20] (and a dedicated frontend team for it :)) [17:38:33] We're overdue a gerrit upgrade too [17:38:38] But as always, priorities :) [17:39:06] [telegram] Yeah, update Mailman first :P [17:41:54] There's a bit of a dependancy tree there [17:42:08] The new UI is allowing us to do cool stuff like: https://gerrit.googlesource.com/plugins/zuul-status/+/refs/heads/master