[04:28:26] Why does the security dropdown only have two options now? "None" and "Private Issue" [07:48:18] "Wikimedia adopts Phabricator, deprecates seven infrastructure tools" session proposal submitted to FOSDEM. We'll see. [07:55:37] qgil: that is a good teasing title :] [07:56:20] hashar, thanks. In fact, I was hesitating between six or seven, because the seventh is Jenkins... :) [07:56:47] but well, counting Bingle, Bugello & co we have more anyway [07:58:41] qgil: well Gerrit / Jenkins are not going to be migrated anytime soon :] [07:59:05] Gerrit before March 2015, according to the current plans [07:59:43] fosdem is in January/February, we should be in the middle of the sh*t by then [08:01:07] i think we will start looking at what needs to be done in January [08:01:28] the whole challenge to me is figuring out how to connect Phabricator and Zuul together [08:01:35] we are committing to a "proof of concept" by the end of this year [08:02:06] (I'm talking about code review, CI is another beast) [08:07:51] qgil: yeah that is what I understood :] [08:08:03] CI is not that hard, though we need Phabricator to emit events [08:08:16] and write some code in Zuul to understand those events [08:16:32] hashar, or maybe we can skip Zuul altogether? See the couple of links I just posted at https://phabricator.wikimedia.org/T31#6406 [08:16:47] potentially [08:17:04] if Phabricator has the utilities to reimplement all the logic we have in Zuul (very very unlikely) [08:17:33] I am 99% sure Harbormaster does not match Zuul feature set :( [08:17:38] hashar, are the features to match documented anywhere? [08:18:04] the most important one is the ability to gate changes together which has a few pages documentation on http://ci.openstack.org/zuul/gating.html [08:18:29] if we have change A , B which are approved with a +2 , we trigger two set of jobs. The first set test A [08:18:37] the second set test A + B together [08:18:54] before A get merged. Zuul speculates that A will pass so it assumes it is fine to test B with A already being merged [08:19:21] if A pass it is merged, and if A + B pass it is then merged. Ie you don't have to wait for A to be merged to run tests for B [08:19:37] (and changes can be in different repositories) [08:20:55] I see. [08:24:02] qgil: but it is surely doable to have Phabricator emit/stream events somehow [08:24:14] and from there adjust Zuul to learn about those events [08:24:25] though that needs a fair amount of code to be written [08:24:29] I'm sure we will find the way [08:31:49] I am sure :] [08:32:02] it will just take a little more time than envisionned [10:01:24] that also depends on when we start [16:09:56] +1 for adding IPv6 :-] [17:52:51] curl -6 -I http://phabricator.wikimedia.org :) [17:52:54] note the "6" [17:57:36] :) [18:39:45] w00t, reply by email working the way I expected :) [18:42:42] you mean, you sent an email and it appeared in phab? :P [18:44:36] also note phab not displays message sources "via email" https://phabricator.wikimedia.org/T475#6427 [18:44:39] kinda neat [18:45:47] Reedy: right :P [18:46:12] chasemp: cool :) [18:46:30] not=now :) [18:47:10] yeppers [19:12:55] twentyafterfour, according to an email notification from T476: "mmodell changed Security from security-bug to none." but I still can't access it. [19:13:39] Then , when I tried to reply via email, I got an interesting reply: [19:13:41] Your email to Phabricator was not processed, because an error occurred while [19:13:41] trying to handle it: [19:13:41] This mail is addressed to an object ("T476"), but that object does not exist. [19:14:36] I'm still thinking whether this is good or not from a security point of view :) [19:39:23] setting an issue to none just means no security settings are applied, it doesn't change the existing security settings [19:39:42] so if you set to none, you can then change the view/edit [19:39:49] we could make a public/none/whatever [19:40:04] but I don't know if that makes any sense at all [19:40:59] Krenair, thanks for your feedback: https://phabricator.wikimedia.org/T516 [19:41:57] chasemp, me neither, but this is something that Security team members need to be aware of [19:42:08] time and practice will tell [20:33:38] qgil: yeah I was pondering whether it should unset the security policy when setting security to none...seems like the sensible thing would be to remove the security policy but it's kinda redundant [20:34:48] qgil: ok now I removed the security policy