Jump to content

IRC office hours/Office hours 2014-11-18

From Meta, a Wikimedia project coordination wiki

Bugzilla-Phabricator migration

[edit]

Log of the first session

[edit]

Time: 16:00-17:00 UTC
Channel: #wikimedia-office
Timestamps are in UTC.

[16:00:20] <qgil> Hello, welcome to this office hour dedicated to the migration from Bugzilla to Phabricator.
[16:00:32] <qgil> Just a short introduction.
[16:00:40] <qgil> Since the approval of the Phabricator RfC at the end of May, andre__, chasemp, twentyafterfour, and qgil have been working mostly full time on this project.
[16:00:53] <qgil> We wouldn't have got here without many other people who has helped commenting, reporting, taking tasks, and resolving them. Big Thank You to all of them!
[16:01:04] <qgil> We had a Phabricator testing instance in Labs (usually remembered as "fab"), and since mid September we have an instance in production: https://phabricator.wikimedia.org
[16:01:17] <qgil> Since then, we have been working in 1001 details related to the migration script, Phabricator, and the bug/task/project management processes.
[16:01:35] <qgil> You can find more details about the project in https://www.mediawiki.org/wiki/Phabricator and https://phabricator.wikimedia.org/tag/phabricator/
[16:01:43] <qgil> We are almost ready to start the Bugzilla migration this Friday at 00:30 UTC. https://www.mediawiki.org/wiki/Phabricator/versus_Bugzilla contains a lot of details.
[16:01:54] <qgil> And a data point: currently there are 812 users registered in Phabricator, which is approximately the number of Bugzilla users active in the past 12 months. We seem to be in a good path!
[16:02:06] <qgil> Alright, let's start the conversation. Who goes first?
[16:03:01] * andre__ wondering who's around
[16:03:20] <qgil> I should have asked that, yes.  :)
[16:03:32] <qgil> o?
[16:03:39] <andre__> ....also note that we have collected lots of information at https://www.mediawiki.org/wiki/Phabricator/Help
[16:03:42] <andre__> Do not tell me that the information is so good and perfect that there are no questions left here :P
[16:03:51] <qgil> (or do tell me)
[16:05:20] <qgil> Anybody here for the Bugzilla-Phabricator migration office hour?
[16:06:14] <andre__> Your questions and input are very welcome.
[16:08:56] <Nikerabbit>  ;)
[16:09:00] <Nikerabbit> somewhat quiet
[16:09:06] <chasemp> (no news is good news)
[16:09:13] <qgil> Ok, this is kind of surprising, but if there are no question / feedback, thhere is no lack of work for us to do.  :)
[16:09:16] <YuviPanda> is there going to be a per-user option to turn on the silly names?
[16:09:35] <qgil> thanks YuviPanda for this question  :)
[16:09:49] <qgil> As far as I know, such option does not exist.
[16:09:50] <YuviPanda> :D
[16:09:53] <YuviPanda>  :(
[16:10:06] <Christopher_WMDE> we have a question about bumping the phabricator version
[16:10:07] <andre__> not that I'm aware of, no
[16:10:09] <YuviPanda> once our migration clears up and things are quieter, I shall file it upstream.
[16:10:12] <valhallasw`cloud> YuviPanda: it's just a translation that can be added
[16:10:19] <valhallasw`cloud> I think?
[16:10:22] <valhallasw`cloud> just like ALL CAPS
[16:10:36] <YuviPanda> valhallasw`cloud: yeah, but have to let people choose it on a per-user level.
[16:11:15] <valhallasw`cloud> it's in your user settings ('language').
[16:11:44] <chasemp> Christopher_WMDE: actually wanted to let you know there was an error on the upgrade related to the sprint libext
[16:11:49] <qgil> Christopher_WMDE, what is the question? chasemp probably can answer it.
[16:12:09] <chasemp> Christopher_WMDE: https://phabricator.wikimedia.org/P78
[16:12:15] <andre__> valhallasw`cloud, oh true, there's the "Translation" dropdown. Thanks for pointing that out!
[16:12:38] <Nikerabbit> How to refer to tickets in Phabricator in gerrit commit messages?
[16:12:53] <Christopher_WMDE> what is the plan for pulling new tagged releases from upstream?
[16:12:54] <qgil> Nikerabbit, Bug T1234
[16:13:05] <Nikerabbit> qgil: also in the footer with colon?
[16:13:21] <qgil> Nikerabbit, https://www.mediawiki.org/wiki/Gerrit/Commit_message_guidelines#Phabricator
[16:13:33] <qgil> yep
[16:13:36] <Nikerabbit> ok
[16:13:52] <Nikerabbit> where does the T come from?
[16:14:03] <qgil> Task
[16:14:09] <chasemp> Christopher_WMDE: kind of a tricky question in that upstream doesn't tag releases, we choose specific upgrade points and then vett and tag a local copy
[16:14:22] <Nikerabbit> ok, it's certainly better than Q :)
[16:14:23] <qgil> (D Diff, F File...)
[16:14:25] <chasemp> and so we are the proprietors of our own "stable"
[16:14:29] <valhallasw`cloud> Nikerabbit: T Task
[16:14:35] * valhallasw`cloud is slow
[16:14:46] <chasemp> but the general plan is to settle on some defined interval, I think monthly is the current intention
[16:15:32] <chasemp> (upstream recommends weekly but that's pretty aggressive)
[16:16:34] <qgil> More questions / comments?
[16:17:00] <Nikerabbit> how old is the Phabricator project itself?
[16:17:28] <andre__> https://en.wikipedia.org/wiki/Phabricator says 2010 :)
[16:17:35] <andre__> for "initial release" though
[16:17:42] <Christopher_WMDE> ok, will upstream pulls be staged in labs for pre-release testing?
[16:18:01] <chasemp> AFAIK it's roughly 7 years old I think
[16:18:25] <chasemp> the main guy was a big part of it inside facebook, they chose to release it, he chose to leave facebook and continue on the phab journey
[16:19:30] <chasemp> Christopher_WMDE: yes vetting in labs is the intention, how exactly to roll it out I'm not sure, maybe 2 week furlough in labs, and 2 weeks behind upstream for our "month"
[16:19:51] <chasemp> https://secure.phabricator.com/book/phabflavor/article/project_history/
[16:20:27] <qgil> "This document is mostly just paragraph after paragraph of self-aggrandizement."oh gosh :D
[16:21:08] <chasemp> There used to be a less official, more light hearted history on stack overflow years ago but I can't find it
[16:22:19] <qgil> Any questions from Bugzilla users that haven't registered to Phabricator yet? (or haven't used it so far?)
[16:24:53] <qgil> Any more questions or feedback?
[16:25:46] <Tobi_WMDE_SW> I'm wondering if we will be able to use the sprint/burndown functionality next week already
[16:26:14] <Tobi_WMDE_SW> since bugzilla will be read-only we would not be able to use scrumbugz for upcoming sprints anymore
[16:26:34] <Tobi_WMDE_SW> so, we are relying on the sprint extension to be ready and deployed
[16:27:06] <andre__> That's a question of resources
[16:27:08] <andre__> Would be cool but I'm slightly afraid we'll pretty be busy with other cleanup tasks already :-/
[16:27:25] <qgil> Christopher_WMDE, do you think the Sprint app is ready?
[16:27:38] <Christopher_WMDE> It is in good shape
[16:27:41] <andre__> https://phab08.wmflabs.org/sprint/view/4/ looks indeed pretty good!
[16:27:51] <Christopher_WMDE> not fully tested, but it works
[16:28:23] <Tobi_WMDE_SW> then it is a question of it being deployed
[16:28:49] <qgil> What we agreed back in September was that it would be ok-ish for you to have one sprint of transition (the current one). Is this assumption still valid?
[16:29:08] <qgil> How often do you run sprints, and when is the current one finishing?
[16:29:20] <andre__> still two weeks?
[16:29:43] <Tobi_WMDE_SW> andre__: still two weeks. yes
[16:29:55] <Tobi_WMDE_SW> next one would start on nov, 25th
[16:30:02] <Tobi_WMDE_SW> so would be nice to have it until then
[16:30:13] <qgil> 25th, this is not two weeks
[16:30:28] <Tobi_WMDE_SW> qgil: it is. last one started on nov 11th
[16:30:43] <Tobi_WMDE_SW> nov 11th - nov 25th == two weeks
[16:30:51] <qgil> ok, undertood now
[16:31:38] <qgil> Christopher_WMDE, Tobi_WMDE_SW chasemp I think it is time to create a task to discuss this deployment specifically
[16:31:43] <Tobi_WMDE_SW> qgil: think we would be fine if we switch back to paper for one
[16:32:15] <qgil> One theoretical possibility would be to run a first sprint with Sprint enabled only to WMDE dev team
[16:32:38] <qgil> but I don't know the actual implications for this i.e. whether other users would see restricted links or something
[16:32:42] <Tobi_WMDE_SW> not sure what is left to do to have it working. the extension is working fine according to Christopher_WMDE and as can bee seen on https://phab08.wmflabs.org/sprint/view/4/
[16:32:54] <qgil> Tobi_WMDE_SW, have you tested it?
[16:33:18] <Tobi_WMDE_SW> qgil: we were playing around on https://phab08.wmflabs.org/sprint/view/4/
[16:33:26] <Tobi_WMDE_SW> https://phab08.wmflabs.org
[16:33:54] <Tobi_WMDE_SW> but the ultimative test would only be with real tasks
[16:36:12] <qgil> Let us discuss it tomorrow in our team meeting. You have done a bunch of work and we want to deploy it as soon as possible, but it's a significant piece of software and I guess that it needs some review before.
[16:36:21] <qgil> chasemp, any opinion?
[16:36:46] <chasemp> Not exactly sure the state of things, has it been deployed in labs using the pod deployment mechanism?
[16:36:59] <Tobi_WMDE_SW> Christopher_WMDE: ? ^^
[16:37:06] <chasemp> that kind of thing should go in a ticket for posterity / async coordination
[16:37:36] <chasemp> but other than those issues and overall if csteipp is ok with it and it doesn't cause other issues it's all good to me
[16:39:10] <qgil> Christopher_WMDE, so what you are saying is that https://phabricator.wikimedia.org/T153 (development) is complete, and now we need to discuss the deployment. Then, let's resolve accordingly, and please create a new task to discuss the next steps.
[16:40:18] <qgil> Maybe you could essay the next sprint in phab-08 while Sprint goes through review, and then we have everything ready for you to start the following one in the production instance (unless you/us have found any blockers)?
[16:41:36] <qgil> I mean, in phab-08 you could create the tasks selected for the sprint as simple redirects to the actual tasks in the production instance, where the actual work would happen.
[16:42:54] <qgil> Tobi_WMDE_SW, Christopher_WMDE , 24th is next Monday, and we are starting the Bugzilla migration in two days. There is literally no time for us to do anything else before your next sprint starts.
[16:43:57] <Tobi_WMDE_SW> qgil: yeah, we're fine with having one sprint the stoneage-way. :D
[16:44:28] <qgil> Well, maybe phab-08 could put you in Bronze Age, but your decision. Thank you for your understanding.
[16:44:47] <qgil> and again, thank you very much to Christopher_WMDE and WMDE for the work done on the Sprint app!
[16:45:17] <qgil> Other questions? Other topics?
[16:45:31] <Tobi_WMDE_SW> let's see how the migration works. would be nice to have the extension deployed for end-nov begin-dec then
[16:46:45] <Tobi_WMDE_SW> qgil et. al: thx a lot! you've done amazing work to bring the ship until here! keep on.
[16:46:59] <qgil> Thank you  :)

Log of the second session

[edit]

Time: 23:00 - 00:00 UTC
Channel: #wikimedia-office
Timestamps are in UTC.

[23:00:21] <qgil> Alright, who is here for the Bugzilla-Phabricator office hour?
[23:00:21] <twentyafterfour> ello
[23:00:25] <andre__> hey everybody!
[23:00:45] <chasemp> I'm here but may be away for a moment
[23:00:51] * bd808 waves
[23:01:14] <greg-g> great, the usual suspects
[23:01:30] <qgil> I'll make a short intro, but please keep waving if you see this. It makes us feel better.  :)
[23:01:42] <qgil> Since the approval of the Phabricator RfC at the end of May, andre__, chasemp, twentyafterfour, and qgil have been working mostly full time on this project.
[23:01:49] * bd808 wonders which of us is Keyser Söze
[23:01:52] <qgil> We wouldn't have got here without many other people who has helped commenting, reporting, taking tasks, and resolving them. Big Thank You to all of them!
[23:02:09] <qgil> We had a Phabricator testing instance in Labs (usually remembered as "fab"), and since mid September we have an instance in production: https://phabricator.wikimedia.org
[23:02:20] <qgil> Since then we have been working in 1001 details related to the migration script, Phabricator, and the bug/task/project management processes.
[23:02:26] <chasemp> twentyafterfour: see -staff chat
[23:02:32] <qgil> You can find more details about the project in https://www.mediawiki.org/wiki/Phabricator and https://phabricator.wikimedia.org/tag/phabricator/
[23:02:37] <chasemp> ppl are getting weirdness from redirector.php
[23:02:47] <qgil> We are almost ready to start the Bugzilla migration this Friday at 00:30 UTC. https://www.mediawiki.org/wiki/Phabricator/versus_Bugzilla contains a lot of details.
[23:03:00] <qgil> And a data point: currently there are 812 users registered in Phabricator, which is approximately the number of Bugzilla users active in the past 12 months. We seem to be in a good path!
[23:03:11] <qgil> Alright, let's start the conversation. Who goes first?
[23:03:11] * quiddity waves
[23:03:25] * spagewmf three thumbs way up
[23:04:14] <andre__> Please ask all the questions you always wanted to ask about Phabricator! :)
[23:04:17] <bd808> What can we expect to see following Day-0? What's the next feature/issue to be tackled?
[23:04:31] <spagewmf> when's the last time you ran the conversion, and do you have a link to a sample converted BZ?
[23:04:44] <qgil> ok, bd808 first
[23:04:54] <qgil> Well, RT goes next
[23:05:06] <quiddity> spagewmf, e.g. = https://bugzillapreview.wmflabs.org/tag/mediawiki-extensions-echo/
[23:05:31] <bd808> qgil: Awesome. That will make things so much nicer
[23:05:35] <qgil> how long will this take right after the Bugzilla migration... chasemp might have an opinion
[23:05:50] <chasemp> how long until RT?
[23:06:02] <andre__> spagewmf: a sample ticket imported from BZ is e.g. https://bugzillapreview.wmflabs.org/T20120
[23:06:06] <andre__> (if "BZ" = ticket)
[23:06:10] <chasemp> if not for thanksgiving I would say following week, but that's a weird week so probably a week+
[23:06:20] <legoktm> o/
[23:06:26] <qgil> Yep, that sounds reasonable.
[23:06:35] <qgil> Also, we will pointing toward https://phabricator.wikimedia.org/tag/project-management/
[23:06:49] <qgil> The next big milestone
[23:07:08] <legoktm> if we have open gerrit patches linking to Bug: ### (where ### is a bugzilla id), will the gerrit-->phab thing, automatically comment on the right maniphest task?
[23:07:10] <qgil> but... before daydreaming (although it's midnight here) let's go back to the Bugzilla step
[23:07:45] <qgil> spagewmf, your question is answered, right?
[23:07:59] <qgil> chasemp, is running migration script tests regularly
[23:08:28] <spagewmf> qgil: Ah, so more recent conversions might have fixes or improvements
[23:08:57] <qgil> It's more about polishing edge cases, right chasemp ?
[23:08:59] <chasemp> minor ones for issue outliers but yes
[23:09:20] <qgil> legoktm, Gerrit links to Bugzilla URLs, which get automatically redirected to the corresponding Phabricator tasks.
[23:09:31] <chasemp> like this one https://bugzilla.wikimedia.org/show_bug.cgi?id=32056
[23:09:38] <chasemp> issue metadata, invalid, no comment 0
[23:09:52] <mutante> i'm going to revert the redirector change
[23:09:55] <chasemp> few of those and they never hit preview so I had to accomodate but that's all special case stuff
[23:10:04] <mutante> since it's reported it causes problems
[23:10:07] <legoktm> qgil: right, but will the gerrit notification bot know how to follow these redirects?
[23:10:14] <legoktm> the one that comments "Change blah blah was merged"
[23:10:30] <qgil> mutante, please use #wikimedia-devtools to report current problems
[23:10:47] * bd808 was wondering about that gerrit notification bot in general
[23:10:55] <andre__> legoktm: I'm not entirely sure but I would not expect that. https://phabricator.wikimedia.org/T169 was the ticket about that
[23:10:56] <qgil> legoktm, ah, I see...
[23:11:02] <andre__> and the code is in https://gerrit.wikimedia.org/r/#/c/167524/ and https://gerrit-review.googlesource.com/#/c/60914/
[23:11:13] <chasemp> qchris wrote a notiification bot to mimic the existing one for gerrit I believe
[23:11:42] <legoktm> basically I don't want to have to manually update ~100 open patches to use a maniphest task # instead of the BZ one
[23:12:06] <andre__> I would NOT expect the Gerrit notification bot to use our redirects to look up that "Bug: 12379" is now "T14379" and comment in Maniphest
[23:12:12] <andre__> but it's an interesting idea
[23:12:18] <spagewmf> legoktm: +2 all your patches before the conversion, problem solved
[23:12:21] <spagewmf>  :)
[23:12:23] <chasemp> is that just a link tho?
[23:12:26] <andre__> yeah. clean up! :P
[23:12:35] <legoktm> spagewmf: I wish!
[23:12:40] <chasemp> if it's a link it should be included in existing logic
[23:12:48] <qgil> legoktm, it's worth writing that problem down in a task
[23:12:57] <legoktm> ok
[23:13:59] <spagewmf> so the day after conversion, Gerrit Notification Bot tries to add "Change 73113 had a related patch set uploaded by Legoktm... Comment" to bugzilla. What happens?
[23:14:26] <andre__> Bugzilla will be readonly. So it won't be able to.
[23:14:42] <legoktm> https://phabricator.wikimedia.org/T1327
[23:15:03] <qgil> This is a problem for the existing open Gerrit changes with a Bugzilla number in the commit message
[23:15:20] <qgil> We might have a solution for this...
[23:15:50] <qgil> But here in an office hour is not the best moment to try to come up with it. Chris is the one that needs to know first.
[23:16:20] <qgil> Thanks legoktm , good catch.
[23:16:27] <andre__> yeah
[23:16:27] <spagewmf> qgil: to refer to Phab tasks in gerrit, should we be adding Bug: T20110 to the commit message? I.e. what does the new qchris bot look for
[23:16:29] <qgil> More questions or comments?
[23:16:34] <legoktm> what happened to the Triagers group?
[23:16:48] <andre__> spagewmf, https://www.mediawiki.org/wiki/Gerrit/Commit_message_guidelines#Phabricator
[23:16:53] <andre__> legoktm, https://phabricator.wikimedia.org/T583
[23:17:05] <legoktm> spagewmf: yeah, Bug: T####
[23:17:11] <spagewmf> \o/
[23:17:12] <James_F> andre__: Why "Bug: T1234" rather than "Task: T1234"?
[23:17:18] <chasemp> no Bug thing is needed I think
[23:17:19] <qgil> spagewmf, sure, if the commit message is edited substituting the Bugzilla report number by Bug: T20110, then it works again. The thing is whether a better solution can be found.
[23:17:21] <chasemp> only T123
[23:17:21] <James_F> andre__: Given that they aren't just bugs.
[23:17:27] <legoktm> > This comment was removed by luser.
[23:17:28] <legoktm>  :|
[23:17:39] <legoktm> are we still going to let people delete their own comments?
[23:17:40] <andre__> James_F, let me see if I can find something about that in https://phabricator.wikimedia.org/T169  :)
[23:17:57] <qgil> Triagers
[23:18:01] <andre__> legoktm: yes, we allow that. Phab admins could restore comments via the command line if needed.
[23:18:17] <legoktm> that seems like a bad indicident just waiting to happen
[23:19:01] <qgil> legoktm, Phabricator doesn't give us another option today, we are discussing this with upstream. I'm sure quiddity or andre__ can find the links
[23:19:06] <quiddity> (that's filed as https://phabricator.wikimedia.org/T1185 "Restrict access to comment removal")
[23:19:31] <qgil> Triagers  :)
[23:19:56] <qgil> Nice idea in theory but a bit controversial and it was causing with problems with permissions to innocent users
[23:20:00] <qgil> we pulled it
[23:20:03] <James_F> andre__: Nothing in T169 that I see?
[23:20:14] <legoktm> okay
[23:20:22] <legoktm> those are all my questions, thanks :)
[23:20:29] <andre__> legoktm, https://phabricator.wikimedia.org/T1185
[23:20:29] <qgil> Now we are discussingthe possibility to bring them back, for a slightly different purpose: limiting access to Batch Edits
[23:21:23] <qgil> (Batch Edits for everybody might be a bad incident just waiting to happen)
[23:21:29] <greg-g> oh, neat
[23:21:41] <qgil> There is also a task for this ( andre__ I keep welcoming your help linking to tasks)  :)
[23:22:06] <andre__> batch edits: https://phabricator.wikimedia.org/T1292
[23:22:43] <andre__> James_F: hmm, then not sure - qchris might know
[23:22:58] <andre__> maybe just not thought about also having "Task: T123", or just "T123"
[23:23:12] <quiddity> re: "Bug: T1234", I think that might regarding differentiating them from BZ links more easily, https://phabricator.wikimedia.org/T857 "[...] Bug 1 -> Task 1; Bug 2007 -> Task 2007")
[23:23:15] <qgil> In general, I believe that the current and potential subtasks under https://phabricator.wikimedia.org/T84 might be an area of activity after the Bugzilla and RT migrations
[23:23:21] <qgil> Your feedback and help there is welcome.
[23:23:37] <qgil> (preventing vandalism)
[23:24:29] <James_F> andre__: Should I file a task?
[23:24:36] <bd808> How high on the post Day-0 list is T493 (Herald)?
[23:24:42] <quiddity> There were a lot of comments in there, about how now is the only chance we have to get that right, and I'm wondering if it would be worth it to keep discussing?
[23:24:55] <qgil> Commit message guidelines thing, this seems to be a minor detail. Convince QChris about the right thing and he probably can do it.  :)
[23:25:01] <chasemp> bd808: pretty much we hope to get it going this weeek, if not then well, soon
[23:25:12] <andre__> James_F: what qgil said ^  :)
[23:25:25] <chasemp> it's a matter of shoring up all the "data leakage" portions of herald like being able to subcribe yourself to tasks you shouldn't be able to
[23:25:27] <James_F> qgil: Kk.
[23:25:34] <bd808> excellent. Thanks for all the hard work on this stuff chasemp and twentyafterfour.
[23:25:41] <James_F> +1
[23:26:03] <spagewmf> I think in gerrit explicit word then Phabricator code is fine, I don't have a preference between Task: or Bug: as the lead-in word.
[23:27:06] <qgil> quiddity, at this point I think your trust on us doing the best thing reasonably possible is the better option at hand. The migration is starting in 48h or so.
[23:27:14] <spagewmf> could gerrit link to a Mock or a Paste or all the other Phabricator letter codes, or is that nonsensical for Gerrit Notification Bot ?
[23:28:20] <qgil> spagewmf, no idea, but I'm not sure the API for non-Maniphest objects is just the same.
[23:28:26] <chasemp> I think there is no lead in needed for linking via gerrit, in phab internals any number with a T lead in is a task
[23:28:27] <chasemp> thus https://gerrit.wikimedia.org/r/#/c/173875/
[23:28:32] <quiddity> qgil, nod. Where pragmatism hits idealism, is the source of almost every disagreement. ;)
[23:28:35] <chasemp> T123 in the commit becomes al ink
[23:29:03] <twentyafterfour> I think our efforts would be best put into switching from gerrit to differential
[23:29:18] <twentyafterfour> rather than more integration between gerrit and phab
[23:29:23] <qgil> chasemp, yes, but currently that doesn't trigger a #Patch-For-Review in phabricator. A details to be polished.
[23:29:32] <chasemp> ah understood then
[23:30:44] <greg-g> twentyafterfour: +1
[23:31:03] <spagewmf> I'm confused, will T20120 anywhere in gerrit comment turn into a link? Meanwhile only something like newline Bug: TNNN should trigger qchris' Phabricator update.
[23:31:53] <qgil> spagewmf, I'm not sure, but this is perhaps the same behavior we had with Bugzilla, just ported to Phabricator.
[23:31:55] <bd808> twentyafterfour, greg-g: I was kind of under the impression that differential was many months away. Is that an old point of view?
[23:32:39] <andre__> spagewmf, hmm, now that you say, the linking indeed might be more aggressive in the commit message in Gerrit compared to linking to BZ
[23:32:50] <qgil> Depends on your notion of "many", but #Project-Management goes first
[23:33:15] <qgil> when it comes to official priorities, that is
[23:33:21] <chasemp> no way for code review this year, and then there is the offsite, etc
[23:33:37] <spagewmf> New Q: What happens to "Bug 12345" references? Looks like See also <bug list> field is added to the Task header, Blocks NNNN field becomes Phabricator's Blocks TXXXX field (Yay), BUT what about straight "blah blah bug 12345" in text?
[23:33:49] <qgil> The Differential path allows more progressive steps than the Bugzilla-Maniphest step, although Continuous Integration is the real bottleneck, probably
[23:34:03] <andre__> re Differential: https://www.mediawiki.org/wiki/Phabricator#Migration_timeline and https://phabricator.wikimedia.org/T560 are related
[23:34:10] <James_F> chasemp: Yeah, but https://gerrit.wikimedia.org/r/#/q/bug:T1296,n,z doesn't work sadly.
[23:34:21] <qgil> spagewmf, you really need to ask all these questions to QChris, and/or in the related task
[23:34:57] <greg-g> bd808: mostly right, yes. But I'm just thinking as long as we have mostly status quo (comments in Phab from gerrit changes) then we should be fine. I'm worried about the gerrit-differential transition from a CI point of view, though :/
[23:35:25] <bd808> greg-g: Maybe we should look at that soon then. :)
[23:35:57] <spagewmf> qgil: OK. I'm not asking for more lines of code, I just haven't found a Phab task converted from a bug with "blah blah bug 12345" in a comment.
[23:36:18] <bd808> greg-g: Like say add dealing with that as a goal in a "make testing better" top5 project
[23:37:33] <James_F> chasemp: I guess it'll become moot when we switch code review over to Phabricator too…
[23:37:35] <qgil> spagewmf, these "bug 1234" mentions in migrated bugs will be plain text at the beginning, but we have a task open to convert them in links after the migration.
[23:37:35] <James_F> chasemp: See in comparison e.g. https://gerrit.wikimedia.org/r/#/q/bug:49998,n,z
[23:37:55] <greg-g> bd808: that balloons it pretty big :/
[23:38:12] <James_F> qgil: Kk.
[23:38:43] <spagewmf> qgil: right, I was just comparing https://bugzilla.wikimedia.org/show_bug.cgi?id=50120#c24 and https://bugzillapreview.wmflabs.org/T20120#1285 It's just plain text.
[23:39:40] <qgil> Converting wiki links in Bugzilla: https://phabricator.wikimedia.org/T875
[23:39:54] <quiddity> spagewmf, converting bugzilla links: https://phabricator.wikimedia.org/T687
[23:40:48] <qgil> Funny, both quiddity and me are right  :)
[23:41:20] <quiddity>  :)
[23:41:38] <qgil> I wonder whether we will apply our Wikimedia/MediaWiki bot intelligence to Phabricator...
[23:41:57] <andre__> one is for our custom en:Foo markup (which is a custom WMF hack in Bugzilla), one is about linkifying "bug 12345" (which is default code in Bugzilla)
[23:42:28] <qgil> then quiddity was more right than me
[23:42:32] <spagewmf> andre__: it would be useful to say for each BZ magic linking feature whether it works in Phabricator and whether it works in converted bugs. E.g. gerrit 12345 is also automagically linked, I think comment 23 is linked.
[23:43:36] <qgil> According to plan, bug 1234 will be linked, but comment 56 not. Mapping comments from both systems is just a mad scientist task.
[23:44:22] <andre__> spagewmf, expect that nothing is linked. ;)
[23:44:23] <qgil> Are there any questions from Bugzilla users tha haven't used Phabricator yet?
[23:45:07] <qgil> This conversation is pretty advanced, and perhaps we have other type of participants in this office hour
[23:45:58] <qgil> I'm actually really really curious about the first impression Phabricator will cause among the wave of new users, also those that didn't use Bugzilla much, or at all.
[23:47:13] <James_F> qgil: Do we have numbers about how many active Bugzilla-ians have already created their linked account on Phabricator?
[23:47:38] <James_F> qgil: 50% of people active in the last month? 10%? We don't know?
[23:47:48] <andre__> James_F, not really "active Bugzilla-ians", but about 800 users have created accounts already
[23:47:56] <andre__> oh, we don't know like that, no
[23:48:02] <andre__> we did not run a match on email addresses
[23:48:04] <qgil> James_F, we don't know exactly, we should map both databases
[23:48:07] <James_F> OK. :-)
[23:48:16] <James_F> qgil: Don't want to delay anything, though, just curious.
[23:48:45] <James_F> If we could ping people that haven't made a linked account but are active in the past month, maybe? Or just let them go through the post-transition sign-up experience instead.
[23:48:47] <qgil> But about 800 users registered to Phabricator, and about 800 users were active in Bugzilla in the past year, so it is reasonable to think that there is a good overlap of active users
[23:49:02] <qgil> I think this is actually pretty good for the launch
[23:49:09] <andre__> +1
[23:49:29] <qgil> "a linked account"?
[23:49:55] * James_F nods.
[23:50:12] <qgil> we have sent two notifications to the 800 active users in the past 12 months
[23:50:14] <James_F> qgil: As in, made a Phabricator account that shares an e-mail address with a Bugzilla account.
[23:50:19] <James_F> Right.
[23:50:20] <qgil> (but I'm not sure I understand your question)
[23:51:30] <qgil> The main goal was to have a critical mass of pre-registered users, so the amount of usernames would win "bzimport" in Phabricator after the launch
[23:51:35] <andre__> I'm not even sure how easily we could access the email address(es) of accounts registered in Phab
[23:51:38] <qgil> I'm confident that we have reached this goal.
[23:51:42] <andre__> ...if we wanted to find out
[23:51:44] <James_F> qgil: Yeah.
[23:51:47] <James_F> Sure.
[23:52:33] <qgil> The remaining ones will quickly see that their names are missing, and will ask/react.
[23:52:34] <quiddity> Could we measure how many Phab users have assigned an email address /at all/ ? (Also how many have specified their mediawiki username)
[23:52:54] <qgil> all Phab users must define an email address...
[23:53:48] <qgil> I guess checking the database we could see how many of them entered via SUL, how many via LDAP.
[23:53:51] <spagewmf> qgil: I agree :) andre__ fine, I added https://www.mediawiki.org/wiki/Phabricator/versus_Bugzilla#Automatic_linking
[23:53:52] <quiddity> ah, ok. (It's been a while since I registered... >.> ) I just see a lot of users with empty profiles, so was wondering if they'd done bare minimum, but missed some important steps.
[23:54:29] <andre__> We can only encourage people to set an avatar or a "real name" (or whatever we call it now)  :-/
[23:54:42] <James_F> andre__: That's a good thing to encourage. :-)
[23:54:48] <andre__> yepp
[23:55:24] <quiddity> I believe it still says "real name" :/ that's https://phabricator.wikimedia.org/T798
[23:56:03] <quiddity> But I meant rather: link to mediawiki userpage, and IRC name.
[23:56:04] * andre__ admits that he does not know the status of each task by heart but is working on that :)
[23:56:04] <qgil> quiddity, that one, and the "Search" one are low hanging fruit that I'm thinking of pushing...
[23:56:26] <qgil> ... it's just that these admins have got so much on their plates, that I hesitate distracting them on minor things.
[23:56:44] <qgil> Then we all get distracted on other minor (and also major) things, but...  :)
[23:56:51] <qgil> 5 minutes
[23:56:54] <qgil> 4
[23:57:44] <qgil> Is there a last question before we close?
[23:58:24] <bd808> Where do we send the thank you beers?
[23:58:38] <quiddity> Send them to January! ;-)
[23:58:42] <qgil> to the MediaWiki Developer Summit, for instance  ;)
[23:58:44] <spagewmf> when does BZ shut down for conversion? https://www.mediawiki.org/wiki/Phabricator/versus_Bugzilla#Timeline doesn't say
[23:59:18] <qgil> spagewmf, "To do, planned for 2014-11-21 00:30 UTC: Bugzilla migration starts"
[23:59:22] <spagewmf> ah, I see above it starts 2014-11-21
[23:59:33] <andre__> (and that's Thu 16:30 in SF)
[23:59:40] <qgil> Is that 48h30'?
[23:59:46] <andre__> yeah
[00:00:03] <qgil> alright, thank you very much for this conversation!
[00:00:04] <spagewmf> No sleep 'til Hammersmith  :)
[00:00:18] <spagewmf> good luck and thanks!