14:09:05 <nils> #startmeeting Modularity Flock 2018 Preparation
14:09:05 <zodbot> Meeting started Tue Jul  3 14:09:05 2018 UTC.
14:09:05 <zodbot> This meeting is logged and archived in a public location.
14:09:05 <zodbot> The chair is nils. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:09:05 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:09:05 <zodbot> The meeting name has been set to 'modularity_flock_2018_preparation'
14:09:12 <nils> #chair sgallagh asamalik
14:09:12 <zodbot> Current chairs: asamalik nils sgallagh
14:09:16 <sgallagh> .hello2
14:09:17 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
14:09:20 <nils> .hello nphilipp
14:09:21 <zodbot> nils: nphilipp 'Nils Philippsen' <nphilipp@redhat.com>
14:09:26 <asamalik> .hello2
14:09:27 <zodbot> asamalik: asamalik 'Adam Samalik' <asamalik@redhat.com>
14:09:58 <nils> and sorry for the confusion, guy
14:10:02 <nils> guys even
14:10:32 <asamalik> https://pagure.io/flock/issue/14  <- that's the one I'd like to talk about the most I guess
14:10:36 <sgallagh> OK, so the main topic here is what we need to prep for the Experts Helpdesk session at Flock
14:10:48 <sgallagh> The Flock organizers have expressed interest in the session, but need details.
14:11:38 <sgallagh> My original vision was basically a hackfest with modularity experts present to help walk people through module creation.
14:11:46 <asamalik> do we wanna do a guided lab? based on the docs we have? I could prepare a completely offline environment for that
14:12:55 <sgallagh> asamalik: Based on the CFP Committee feedback, that might be a good idea, but I think it should be a separate session or an intro section to the helpdesk
14:13:22 <sgallagh> asamalik: How would we have an offline environment? Local module builds?
14:13:39 <asamalik> sgallagh: do you envision a "booth" kind of thing where people would walk in any time? or a scheduled slot of 1 or 2 hours?
14:13:41 <sgallagh> What would we do, prep a virtual machine with all of the cache set up?
14:14:03 <asamalik> sgallagh: the offline environment would be for local builds, yes, just to demonstrate the basic mechanics
14:14:06 <sgallagh> I don't think a booth-style would really work
14:14:45 <sgallagh> I want to aim for "Interested package maintainers come to the session planning to add alternative streams to Fedora for their package"
14:15:03 <sgallagh> We could then do a basic intro lab and then help them apply that to their real problem
14:15:33 <asamalik> sgallagh: do we trust the hotel wifi that much?
14:16:09 <sgallagh> Well, we don't really need the WiFI very much if we rely on the Fedora infra
14:16:14 <asamalik> I definitely like the idea, and I believe it would be great in an ideal world of working hotel wifis :P
14:16:56 <sgallagh> Of course, we could also ask the organizers if it would be possible to get a wired connection in the room
14:17:10 <asamalik> that could be an option
14:17:42 <sgallagh> I'm sure we could find a spare AP around for a private network for just this session as well (for those without physical ethernet ports)
14:18:03 * sgallagh takes that question to the organizers
14:18:58 <asamalik> or we could do the guided lab —so everyone is on the same page, and then dive deep for specific questions or scenarios..
14:19:04 <asamalik> and record that and use that for our docs
14:19:37 <sgallagh> I think we should plan to update our docs with the results of this session no matter what
14:19:42 <sgallagh> Notes should be taken
14:19:48 <asamalik> yes
14:20:15 <asamalik> even if we don't have answers, we can always create an issue or something so we know what to find answers for
14:20:32 <asamalik> ^^ that could be one of the goals we could mention in the proposal
14:20:47 <sgallagh> ack
14:21:56 <asamalik> thinking about the offline lab... can we even do local module builds with RPM sources on a local machine?
14:22:01 <sgallagh> For the moment, let's assume that we can have hardwired access and plan around that.
14:22:23 <sgallagh> asamalik: The MBS caches everything it needs the first time it's run
14:22:35 <sgallagh> So if we build up a VM image that has it pre-cached, I *think* we'll be okay.
14:22:47 <sgallagh> It might still need access to PDC, I suppose.
14:23:10 <sgallagh> I think we need to just state that "reliable, dedicated network" is a requirement for this session.
14:23:12 <asamalik> sgallagh: what if I wanna create a completely new stream with new packages that never been in Fedora before.. do I need to push them to dist-git first?
14:23:23 <asamalik> I just realized that's something I'm missing in the docs
14:23:28 <asamalik> and I'm not even sure we can do that
14:24:01 <sgallagh> asamalik: https://github.com/fedora-modularity/libmodulemd/blob/master/spec.v2.yaml#L258
14:24:41 <sgallagh> I don't think I've tested that feature in a year, but it worked way back
14:24:47 <asamalik> sgallagh: that's the theory... also, not very nice if there's many packages :( but that aside, does it even work?
14:25:11 <asamalik> I feel like we might wanna test this before submitting the proposal :D
14:25:59 <sgallagh> Testing before putting it into production? Where's the fun in that?
14:26:42 <sgallagh> asamalik: What does the COPR modularity support do here?
14:27:01 * sgallagh has never used it
14:27:06 <asamalik> 🔥 ¯\_(ツ)_/¯ 🔥
14:27:37 <sgallagh> asamalik: I think we should propose the session as we've discussed it and then work out the details as we prepare materials.
14:27:53 <asamalik> sgallagh: as far as I know, copr has its own implementation of the MBS functionality... I would have to check again
14:28:06 <sgallagh> If it means begging mprahl and jkaluza to fix a few things, I'd like to believe they'd help :)
14:28:40 <asamalik> sgallagh: I feel like hoping and relying on a good internet connection is optimistic :)
14:29:40 <sgallagh> asamalik: I think it's possible if it's planned for ahead of time.
14:30:03 <sgallagh> Conference wifi in general isn't great, but we can usually arrange for a dedicated line if we need one for a sessio.
14:30:34 <sgallagh> We did okay at Flock Cape Cod.
14:31:43 <asamalik> sgallagh: I'm a little worried, but if you believe we'll have a connection then I'm +1
14:32:03 <sgallagh> asamalik: As I said, I'd put this as a non-negotiable prerequisite.
14:32:10 <asamalik> we can always make more proposals, one offline, one online, and let them accept/reject based on whatever is available
14:32:14 <sgallagh> If the conference center cannot provide, we scrap this plan.
14:32:30 <asamalik> yeah
14:32:33 <sgallagh> I'm pretty convinced that offline isn't possible
14:33:06 <asamalik> sgallagh: everything is possible, you just need to like shell scripts and pain
14:34:52 <sgallagh> I don't like pain. It hurts me.
14:35:44 <sgallagh> To summarize, I think what we're proposing is as follows: << EOF
14:37:02 * asamalik waits on IO
14:37:06 <sgallagh> 1) This session is contingent upon having reliable, dedicated internet access separate from the standard conference wifi.
14:37:06 <sgallagh> 2) We will start with a lab session (time estimate?) to walk through the basic processes of creating and building a module. The environment will be a pre-created VM image.
14:37:44 <sgallagh> 3) After this, we will leave open at least one hour (preferably two) for helping packagers work on their real-world modules and hopefully get them into F29.
14:37:50 <sgallagh> EOF
14:39:23 <sgallagh> asamalik: Does that sound like a valid summary?
14:39:31 <asamalik> would no internet connection mean no modularity lab? or just not having a part of it?
14:40:03 <sgallagh> If we can't have internet, then the only thing I think we can do is a talk with recorded demos.
14:40:21 <sgallagh> We wouldn't be able to really work on their real-world modules either
14:40:39 <nils> Does that need a few words about minimum bandwith reqs?
14:41:02 <sgallagh> nils: I think our bandwidth reqs are pretty low, it's the reliability that's important.
14:41:13 <nils> Mhm.
14:41:38 <sgallagh> But yeah, maybe we should say "minimum 3Mbps"?
14:42:03 <sgallagh> (so we don't end up getting handed a 768k DSL connection)
14:42:33 <asamalik> I'm not really convinced we need to do the real modules there.. we might want to answer the questions about them.. but there will be a lot of time-consuming work packagers could do on their own
14:42:42 <asamalik> and we could use that time more effectively
14:43:04 <sgallagh> asamalik: Well, how are they going to know what questions they have if they aren't trying it right there?
14:43:15 <sgallagh> The advantage for them is having that tight feedback loop with us
14:44:17 <asamalik> sgallagh: my thinking is that we could just cover the basics so people would be confident about:
14:44:22 <asamalik> 1) adding packages, defining modules
14:44:36 <asamalik> 2) building and debugging modules
14:44:49 <asamalik> 3) the basic mechanics such as streams, build order, profiles, and I don't know what
14:44:58 <asamalik> 4) where and how to find/ask for more info
14:45:07 <sgallagh> The real "hard problem" in modules is splitting them and managing dependencies/MSE
14:45:19 <asamalik> so the goal of the workshop won't be new modules created, but people knowing what to do or how to find help
14:45:31 <rcrit> and validating the YAML, right now you get a yes/no only, not a "it is broken on line 12"
14:45:37 <asamalik> sgallagh: so that's what we could discuss there
14:46:26 <sgallagh> rcrit: nils has a linter in progress that will be better about htat.
14:46:32 <rcrit> cool
14:46:43 <sgallagh> But the "no" case on libmodulemd 1.4.1+ should be printing the reason at least.
14:47:17 <asamalik> kind of like a driving school.. we won't go shopping for groceries nor pick up the kids from school, but teach the person how to do that on their own
14:47:27 <asamalik> does that make sense?
14:47:55 <asamalik> sgallagh: and if so, is it something you think it's useful? or do you think that doing the modules would be better?
14:48:09 <asamalik> I can see value in both
14:48:48 <sgallagh> asamalik: My driving instructor once had me pick up his dry-cleaning...
14:49:06 <asamalik> sgallagh: haha I've been to places, yeah :)
14:49:35 <asamalik> but do we want to be those instructors? :D
14:50:10 <sgallagh> asamalik: I think promising a completed module is unlikely to be achievable.
14:50:28 <asamalik> thinking about it again... I'm more and more convinced about an alternative: a kind of semi-hands-on BoF about building modules
14:50:36 <sgallagh> But I think the goal should be to work on them and get them to a state where the person can either finish it themselves or at least engage with us in this channel from a position of basic understanding
14:51:27 <sgallagh> asamalik: OK, I've just been told that reliable internet is unlikely :-(
14:51:39 <asamalik> sgallagh: I was expecting that :(
14:52:19 <asamalik> so what if people came with a module in their mind... and we would collectively discuss best practises/architecture/workflows of making those
14:52:42 <sgallagh> That feels more like the "booth" idea now
14:53:01 <asamalik> or a semi-hands-on BoF
14:53:05 * asamalik likes the name now
14:53:09 <sgallagh> Maybe we scrap this as a session and ask for a dedicated space that we have someone staffing on and off throughout the event
14:53:51 <asamalik> sgallagh: the benefit of a scheduled session would be the collective discussion, so more people would learn from each scenario
14:54:48 <asamalik> sgallagh: like for example a FreeIPA module.. that's not gonna be one huge ugly thing.. diving in details there could be useful for more people... or another one: a pile of nodejs libraries for each nodejs stream
14:54:54 <asamalik> discovering common patterns maybe?
14:55:28 <asamalik> and taking notes and documenting later
14:56:54 <asamalik> but I like the booth option too... just will we have enough people to staff it for the whole event? because I imagine we'll want to participate in other stuff, too
14:57:07 <sgallagh> asamalik: I didn't say the whole event.
14:57:29 <sgallagh> But maybe we pick two one-hour shifts per day and share the load.
14:57:40 <asamalik> also, having all the people there at once could yield better answers
14:57:47 <sgallagh> mm
14:58:16 <asamalik> it's not just us telling people what to do.. it's all of us working together, coming up with (possibly) new patterns..
14:58:27 <asamalik> as I imagine it it at least
14:59:55 <asamalik> Modularity semi-hands-on BoF :D I would even call it that
15:00:25 <sgallagh> asamalik: I don't really have any better ideas :-/
15:01:23 <asamalik> yeah me neither... I did an offline demo from a user's perspective and that was easy, but thinking about the build infrastructure, it's a much more complex thing to do
15:02:06 <sgallagh> Yeah, I think an "offline" demo would require mirroring the build infrastructure in a dozen VMs :)
15:02:31 <asamalik> sgallagh: can we gat a dozen laptops and are we willing to do it locally? :)
15:02:34 * asamalik might be crazy
15:03:22 <asamalik> but probably not... even though I think that could be cool, it would be significantly more work with not that bigger impact...
15:03:27 <sgallagh> asamalik: Assume no
15:03:54 <sgallagh> We had a hell of a time getting even six laptops to use for video recording at Cape Cod
15:04:43 <asamalik> so my proposal is:
15:05:04 <asamalik> "Modularity semi-hands-on BoF" (I'm not sold on this exact name, though)
15:05:30 <asamalik> covering the very basics in the beginning, maybe a recorded demo or something — just to bring everyone on the same page
15:05:46 <asamalik> and then having people introduce their scenarios
15:05:49 <asamalik> maybe vote on them?
15:05:51 <sgallagh> asamalik: Can you turn that into a more detailed proposal and email it out?
15:05:53 <asamalik> and solving them
15:05:54 <asamalik> EOF
15:06:06 <sgallagh> We're kind of over time today.
15:07:17 <asamalik> sgallagh: sure, no problem
15:07:24 <asamalik> sgallagh: so you like that idea?
15:07:36 <asamalik> or at least, you think it's the best we have?
15:07:48 <sgallagh> asamalik: I like it better than inviting people to yell at the WiFI
15:08:00 <sgallagh> But yeah, I suppose it's the best we can do in the situation
15:08:03 <asamalik> haha agree
15:08:22 <sgallagh> However, I think I'd like to take our original idea and pitching a FAD for it
15:08:30 <sgallagh> s/pitching/pitch/
15:08:35 <asamalik> sgallagh: do you want me to add it to the pagure issue?
15:08:42 <sgallagh> asamalik: Please do
15:08:52 <asamalik> sgallagh: or do you still wanna try to do the first proposal, in case something crazy happens and we get a connection?
15:09:01 <asamalik> even though it's unlikely
15:10:19 <sgallagh> asamalik: No, bex pretty much said it wasn't going to happen.
15:10:29 <sgallagh> They have no budget to spare on internet
15:10:44 <asamalik> sgallagh: ack, I'll add it to the existing issue
15:12:34 <sgallagh> Thanks
15:12:58 <asamalik> #info We've agreed we'll do a kind of "semi-hands-on BoF" for Modularity, covering the very basics to get everyone on the same page, and then going through individual scenarios people prepare and discussing them, answering questions, discovering new patterns, etc. All questions will be recorded and used to enhance our docs.
15:13:13 <asamalik> nils: ok, I think you can kill it with fire now :D
15:13:17 <asamalik> sgallagh: np!
15:13:28 <nils> #endmeeting