[05:11:49] legoktm: hey. got some minutes ? Need help with your 'Uh, why not do an actual database rollback?' in https://gerrit.wikimedia.org/r/#/c/304692/17/includes/NewsletterStore.php [05:12:48] I am just wondering how the actuall rollback would really work, and if you got some examples in some othere extensions/core which do that, please share [05:14:28] is it like we create a savepoint, and then rollback to that one (like savepoint before we created the newsletter, and then rollback to that one) ? probably not as there are chances lot other operations would've happend in the newsletter table by that time [07:16:58] legoktm: ... or is there a way to run a rollback to a contenthandler::edit ? I think that should be the one we are looking for, as here is the call happens in Line 187, in https://gerrit.wikimedia.org/r/#/c/304692/17/includes/NewsletterEditPage.php [07:17:28] right after the Newsletter::edit( ) is called with our new newsletter params, and not an add to database, which could be rollbacked [07:29:16] alright. so hoo tells me that "hoo> That means you do IDatabase::startAtomic … everything that should be together … IDatabase::endAtomic" [10:04:40] addshore: when you are around, we got https://gerrit.wikimedia.org/r/#/c/304692/ to review! [16:34:21] andre__: you around ? [16:35:23] finding easy bugs in phab has become difficult, considering that some of experts (fresh more than 5-6 patchsets people) are fixing them for fun. I was talking to Abdeali about this, and he told me that in his org (coala), they do not allow someone with more than say 5 patchsets to fix something which is tagged easy. [16:35:50] and what more -- if an 'expert' submit a patch, he/she is asked to abandon it, and that one is given as an example for someone to rework on and fix the same thing [17:00:22] seems to me we should figure out how to motivate the creation of more "easy" task writeups, rather than punishing experts for something tagged "easy" [17:02:07] robla: true that one too. We had people pulling in things from Easy tag for our last hackathon, and many of them had just one line descriptions. At another hackathon, I came across a set of easy related ones whcih was never tagged 'Easy' [17:02:24] related-similar tasks with some common parent is very helpful for hackathons! [17:02:46] like https://phabricator.wikimedia.org/T145728 [17:03:16] if an expert can fix one among those, getting the others in line would be very easy for a new-comer [17:04:30] oooh, right. Yeah, if we build the system correctly, experts will be able to trust that newcomers will follow their example [17:05:30] true. Take a look at hte fetchContent() thing in that above parent task - we had one of them already fixed by (Reedy, I guess) and the participants could actually take a look at it, and fix all the remaining in fetchContent [17:06:28] (anyway, very difficult to get easy related stuff like this) [17:14:35] tonythomas: about finding ways after having worked on some easy tasks, also see https://phabricator.wikimedia.org/T131706 [17:15:11] and indeed, 'cloneable' cleanup/tech-debt tasks can be a very good way! [17:15:19] * andre__ has to leave this office soon [17:15:27] andre__: true, which might have related! [17:16:15] andre__: https://phabricator.wikimedia.org/T131706 --> I can contribute to that one. [17:16:35] the unclear task description scares me to asssign it to myself though. [17:18:10] haha [17:18:15] {just do it} [17:18:29] tonythomas: I can always ask you in a few months how long you still want to lick the cookie ;) [17:18:30] alright. [17:19:50] done.