Thursday, 2017-07-20

*** emagana has quit IRC00:13
*** Rockyg has joined #openstack-tc04:07
*** lbragstad has quit IRC04:25
*** Rockyg has quit IRC07:24
*** cdent has joined #openstack-tc09:49
*** sdague has joined #openstack-tc10:34
*** cdent has quit IRC13:10
*** lbragstad has joined #openstack-tc13:24
*** marst has joined #openstack-tc14:41
*** cdent has joined #openstack-tc14:56
* smcginnis flips sign to open15:00
fungiooh! office times15:00
smcginnisfungi: Do you attend every office hour? :)15:01
fungi#define "attend"15:02
cdentfungi never sleeps, always at work creeping through the earth, spreading spores15:03
fungionly the ones where i'm not in standby recharge mode15:03
smcginniscdent: Hah!15:03
smcginnisfungi: I obviously need standby recharge more than you do. ;)15:03
fungii actually don't generally attend the "tuesday" office hour because i would either need to sleep in the middle of the day or be up for too long between it and the "wednesday" one15:04
fungidoable, just not convenient15:04
fungieverybody excited about the new sigs framework?15:06
dtroyerTBH I'm not sure how it is more than shuffling things around, much like the middle manager rotation when a company director changes.  Not against it but the value isn't obvious to me.  Also, maybe I just haven't read everything yet (it's been a hit-and-miss summer for me)15:08
dtroyerthat said, branding does make a difference in many places, that may be my blind spot15:09
cdentI worry about that a bit too (the shuffling) dtroyer15:09
fungiit's more of a psychological experiment, to see if we can convince "operators," "users" and "developers" that they can actually work together in groups to get something done15:09
cdentbut yeah, what fungi said15:09
fungiit doesn't actually enable anything governance-wise that wasn't already possible, but we've got these community silos we've been so far relatively unsuccessful at breaking down15:10
dtroyersure, but it seemed more like people interpreted the 'who can be part of this' based on what sponsoring body it came from and not what it was called15:10
johnthetubaguy+1 for removing the barriers, lots of folks "didn't feel invited" to some conversations in the past15:11
dtroyerI'm all for removing those, my doubt is that the identified branding was in fact the issue15:11
smcginnisYeah, I'm mixed. Naming can be important as far as perception goes, so maybe it will make a difference.15:11
johnthetubaguyI guess that's the trick, making folks feel invited to the conversation15:11
lbragstadjohnthetubaguy: question about the baremetal/vm group - would you view that as a sig of some sort?15:12
dtroyeralso, it feels a little bandwagon-ish to cite k8s use of the term15:12
johnthetubaguylbragstad: I totally would15:12
lbragstadjohnthetubaguy: ++15:12
dtroyerbut maybe that'll change the perception15:12
fungiand it really is more psychology/sociology than anything. i mean we already have a lot of people who are operator-developers or user-developers. the ones who feel disenfranchised are, ironically, the ones not just going ahead and getting involved across those silos15:12
cdentdo we know to what extent any of the people not getting involved are (not) doing so because they feel it will be wasted effort?15:13
fungino clue. probably hard to figure that out without polling (and even then responses may not be genuine or representative)15:14
* cdent nods15:14
lbragstadcdent: do you mean across silos?15:14
cdentlbragstad: just generally not stepping up to participate in the sorts of things that sigs, working groups, user groups, etc are wont to do15:15
fungii mean, there are for sure some people who feel like they should be entitled to getting things done for them just because they ask for it. those people are used to vendor/customer relationships and are having a hard time mapping their mental model of software development to ours15:15
smcginnisI guess that's the thing. Whether it's called siG or WG, it needs to have something to show it actually gets things done. Otherwise not a lot of incentive to join.15:15
smcginnisfungi: I think that's a good point.15:16
smcginnisIt's not some sort of "customer advisory board" to a vendor.15:16
cdentit’s not, but maybe some of them could or should be?15:16
fungito a vendor, probably yes15:17
cdent(I wouldn’t like that, I would think it is icky, but it is often how specialized things get done)15:17
fungibut i don't want to go assuming that they're all (or even a majority) of the ones not getting involved across those perceived relationship boundaries15:17
smcginniscdent: That's fine. But it always comes down to - who is going to actually do the work.15:18
johnthetubaguyI keep seeing two cases (1) stuff devs are doing already, but could do with timely feedback, unsure how to get it (2) stuff no one is working on, but users want15:18
lbragstadsmcginnis: ++15:18
fungiwell, if they're advising a vendor, then it's the vendor who needs to step up and get involved in that case. which is fine, we have plenty of those transitive user relationships already15:18
cdentI’m certainly aware that there are _plenty_ of people who don’t engage in feature discussions or planning because of the notion that the idea will get near nova and then die (because of nova’s other priorities) so why bother?15:18
cdentunless we can change that constriction, not much can happy15:19
cdenthappen, but happy will do too15:19
cdentsmcginnis: of it is always down to who will do the work, which is why I keep asserting that the board needs to be more involved in some of these thngs. Not becaus I hate or don’t believe in open source, but because there’s no such thing as a free lunch.15:20
lbragstadregardless of how the notion dies - the steps taken afterwords seem pretty important15:20
cdents/of/of course/15:20
johnthetubaguycdent: what would the board do, recommend priorities?15:21
smcginniscdent: Maybe the reverse of our top 5 help wanted list?15:21
fungii'm not opposed to appeals to the board, it's just they rarely have any positive mipact (cf. my anecdote on the sigs ml about the interop effort)15:21
smcginnisA list of things needed that sponsor companies should be expected to dedicate some resources to addressing?15:21
johnthetubaguyso the VM & BM idea (before I went into a hole because I don't have a job anymore) was really to try and get those kinds of lists15:22
cdentjohnthetubaguy: the eventual outcome I would hope for (and I know this is likely a total pipe dream) is increased investment so those of us already over-invested do not continue to be overworked and constantly disappointing people15:22
cdent(investment by the corporate board members)15:22
johnthetubaguycdent: ah, that's back to the getting permission to work on things, got you.15:22
fungii think in many cases the board members are too far removed from the teams at their organizations doing actual development work upstream, or end up instructing managers to create teams of new hires, interns or otherwise uninitiated newcomers to try and tackle something15:23
cdentno so much getting permission. it’s more about: getting out of constant crisis mode.15:23
johnthetubaguycdent: yeah agreed, that often requires permission of some kind to ignore some stuff15:23
cdentI’m not enough of a capitalist to not think the openstack community isn’t being taken for a ride15:24
dtroyerI'd say (from observation) that some of the board members are way disconnected from what their teams are doing, and often look at it like a regular software development process too.  "Move N heads from project A to project B next quarter."15:24
cdentwe said in boston that we needed to do another round of “this is how openstack development works” but I think we got distracgted15:24
lbragstadcdent: rookie question - but when did that happen previously?15:25
cdentlbragstad: I don’t know, ttx just mentioned it as an “again” when he said it in boston (iirc)15:25
dtroyerlbragstad: maybe back when the community was small enough that we sat in a room and knew where we all stood.  momentum has carried us this far15:26
* lbragstad nods15:26
fungia "bottom up" approach is probably needed for critical efforts, where the real request we should make to the board is extra latitude for the current upstream developers to prioritize and tackle problems we define even if somewhat disconnected from the organization's bottom line, along with efforts to bring new contributors into the community at a comfortable and measured pace so they can be onboarded15:26
fungiwith simpler efforts before jumping onto the harder ones15:26
dtroyerfungi: there you go, making sense again15:27
fungibecause asking the board members to "help us implement feature x" isn't really going to head in a productive direction most of the time15:28
cdentI fear that it has to be both, otherwise why not go elsehwere?15:29
fungithe hard things which are getting tackled successfully are mostly being done by the long-time entrenched and enfranchised community members who know how to make things happen, but there's not enough of them and they're already spread thin15:29
fungiand i think we probably have plenty more of them hiding out in the community, currently loaded down by their employers with other directives15:30
fungiwho would jump on these sorts of problems if they had permission to spend their time on them15:31
johnthetubaguyI was so hoping the PWG was going to help with this, but that dream died a few years back15:31
cdentwe also have plenty of people who are long time contributors who are not officially allowed to contribute as much any more and who are being squashed under too many responsibilities (not speaking of myself here) and if we let that continue they will burn out15:32
fungicdent: i completely agree there (also not me thankfully, but i know of lots)15:33
cdentso to some extent I think we have a labor relations problems15:34
lbragstadIMO - fixing that feels like we need to alter the decisions made by the sponsoring company15:35
fungiwell, in some cases it's people moving into other roles (management, different departments) by choice but still trying to stay afloat with leadership responsibilities in the community even though they have very limited personal time to do so15:37
fungiand this is where bringing up more leaders from within the community is really needed15:38
fungiso those people can finally feel comfortable "letting go"15:38
*** Rockyg has joined #openstack-tc15:38
cdentof course, another aspect of the problem, is that casual contribution is not favored in the projects that are large and busy. if you want to get your code merged, you often have to show up  day in and day out. that’s not a very “open” way to do source.15:39
johnthetubaguyI think someone told me that the linux foundation does a better job of educating companies15:40
johnthetubaguydoes anyone know if that's true, and how it works?15:40
fungicdent: i think it's a problem of volume. any project with high throughput is not especially friendly to casual contributions. ever tried to get a patch into the linux kernel?15:41
sdaguecdent: maybe, I think that's pretty standard with any busy project though. If I want to do anything past trivial bits in home-assistant, I need to show up pretty regularly and lobby on changes (and about 1/2 of those have been rejected on principle)15:41
smcginnisjohnthetubaguy: I know someone from the foundation came out on site to talk to some folks when I was at Dell.15:41
cdentjohnthetubaguy: I remember that from the conversation that happened in boston about doing another batch of education, someone suggested using the linux foundation as a model15:41
smcginnisMaybe they just do more outreach?15:41
sdaguefungi: yeh, the kernel is way on the far side of it, it's almost it's own anti pattern.15:41
cdentfungi: only in like ’94 or ’94 or so, was pretty easy then ;)15:41
fungicdent: indeed ;)15:42
cdentsdague: at least you got rejected rather than ignored?15:42
sdaguecdent: some times15:42
sdaguehonestly, I get ignored on small projects pretty regularly as well15:42
fungii had a significant regression fix for bandersnatch sit for 6+ months without acknowledgement15:43
sdagueI've had outstanding critical fixes on the calendar component in drupal for years that I had to just give up on15:43
sdaguebut, anyway, that's kind of neither here nor there, other than stating this is not a unique problem to openstack15:44
fungiright. we should not be complacent about the challenges to getting attention on patches in openstack, but we still need to keep in mind that we're far from unusual in the open source/free software world where that goes15:45
cdentsdague: yes, agree, but there is a critical difference in that some portion of that work you’ve had ignored or rejected was _your_ work, not work you’ve done as labor for some other entity. And some porition is too.15:45
sdaguejohnthetubaguy: my experience is a lot of education was done directly by the LF, as well as match making members to get people from platinums to go and talk to newer member companies15:50
sdagueyou'd need LF staff to really break down how much time they spend doing that15:50
cdentI think from the standpoint of _many_ contributors, participating in openstack is a job, and a hard one. And for many companies that try to contribute effectively, it is often harder than they think it should be. There are many reasons for that, and many of them are common across similar environments. Does being common make something normal or good?15:50
cdentDuring the mirantis constrictions several people reported “thank god, I don’t have to work upstream anymore”15:51
sdaguecdent: no, I'm just trying to figure out what levers we have to make things better at what cost15:51
sdagueand I've yet to see effective levelers proposed for "make companies invest in open source more and everyone love doing it"15:52
sdaguelevers that is15:52
cdentI can think of one: Help to remind each other to not try so damn hard all the time.15:53
sdaguecdent: ok, would that have solved the mirantis situation?15:53
cdentthe mirantis situation is a secondary effect of the people who already do try too hard demonstrating that the only legitimate effort is extreme effort15:54
cdenthell, people have written blogs that the only way to be successful in openstack upstream contribution is by doubling down on effort15:55
* dims peeks15:56
cdentif enough people have bought into that, then the rentiers aren’t going to bother investing15:56
johnthetubaguyso I guess the bit I struggling with is, what do we ask the board / foundation to do more of?15:58
lbragstadi think that is going to depend on the thing sdague touched on15:58
johnthetubaguyis it talking about investing in open source, how it pays off, go through good examples that have worked well for folks, etc?15:58
lbragstadwhich might require getting more info from the LF?15:58
cdentthat’s at least a good starting point16:00
fungii'm curious how the size of the lf staff compares to osf (we only have a couple dozen people at the moment)16:01
fungilooks like their leadership roster alone is more people that osf employs in total16:02
fungi(lf directors and executives)16:03
cdentwell they are _linux_ after all16:03
fungiheh, the "linux foundation" is so not the linux kernel project, aside from their officially employing linus16:03
fungibasically "open source development labs" merged with the "free standards group" and then renamed their trade organization16:05
cdentsure, but the branding matters: they get to think of themselves as being central to lots of stuff16:05
fungimostly to capitalize on the popularity of the name "linux" from what i can tell16:05
cdentopenstack may be huge but it is not that huge16:05
fungiif memory serves, as a community we surpassed the commit throughput of the linux kernel some years ago. not that it's really a good metric or anything16:07
cdentunderlying everything I’ve been saying on this topic is this one thought: I don’t think we can ask much more of existing people unless they are able to stop doin something they are already doing.16:07
sdaguefungi: only if you count things kind of funny16:07
sdaguecdent: yeh, I think that is spot on. Existing folks are definitely already at carrying capacity16:08
cdentand that carrying capacity is why I think the problem lies with the orgs that provide people. They need to allow people to change what they are doing and/or provide other people.16:09
fungihowever, some existing folks may be willing to shift gears and drop what they were working on to prioritize other things, and more of them could but for having the permission of their employers to do so16:09
johnthetubaguythat is the big catch here, the "free" capacity to train new folks is not really there right now, not without loosing other things16:09
johnthetubaguyyeah, thats the permission bit I was thinking about before16:09
sdagueso, levers...16:10
sdague1) education back into orgs about open source (ala LF model), which has a prereq of understanding what they do beyond just the hand wavy that we have16:10
sdague2) figure out things we'd recommend just get dumped?16:11
cdent3) put it (once we figure what it is) on the official agenda for the board/tc/uc thing at ptg (we need to insure the topic is visible even if we haven’t figured out solutions)16:14
cdent3.5) figure out what it is16:14
fungiwe do have osf staff who do member company outreach and education, but only a few. i expect lf has a good sized department, so increased education may involve board approval for staffing up a dedicated department or engaging a third party to do that for us (and i'd be hesitant about the latter conveying the right messages)16:15
sdaguefungi: it might, they also do some match making. I definitely gave talks as part of a member company to other member companies about open source development. Because the foundation did the match making and set it up, it meant you could talk across vendor lines that you don't normal get to16:16
cdentfungi: do you think a dedicated department can do it? would it not be something, at least initially, that we would need to do?16:16
johnthetubaguysdague: +1 for match making16:17
fungisdague: sure, if there are already a sufficient number of our member companies who "get it" then that could work, but presumably that needs to be at or above 50% to pair them up16:18
fungicdent: by "we" you mean... the tc? the community at large?16:18
cdentfungi: I was thinking “tc” as “representative of the developers”16:19
fungiso basically have tc members visit various companies and give talks?16:20
cdentit’s not clear to me at this stage who or what needs to be talked to. as you (I think it was you) the gap between the people making practical active decisions about resources and plans are the people are board is often huge16:22
johnthetubaguyit feels like the business folks need some exec to reassure them its worth investing, then the dev thing follows up, but I could be seeing this wrong16:22
johnthetubaguyyeah, basically what cdent said16:22
cdentjohnthetubaguy: I’m glad your able to somewhat translate my horrible typing.16:23
* cdent sighs and dies16:23
fungisome (usually larger) companies have a open source/free software liaison position whose job it is to try and at least do internal education16:24
fungii've also seen a few where they just put that person over a division of the company where all upstream involvement/contribution gets assigned16:25
sdagueso... does that mean we're mostly still in a research phase? And we don't have enough data to have an idea what levers might be useful?16:30
johnthetubaguysdague: I think that's what I am hearing16:30
sdagueso... then what questions do we need answers to discover those things?16:31
johnthetubaguywell I like the question you posed about find out what the LF actually does to help16:32
johnthetubaguyI guess there is the bit about discovering what is already done today for OpenStack, I have indirectly heard of the foundation reaching out to folks and helping16:33
cdentdo we have a grasp on the problem, enough that we can make a succinct statement about what it is?16:33
johnthetubaguyprobably need some lists of what needs help, and the folks16:34
johnthetubaguycdent: that too... we need that to raise it in the board meeting16:34
cdentbecause I think we have some differences of opinion on at what layer of the problem(s) we’re trying to attack things16:34
cdentor even discuss things16:34
johnthetubaguyfeels like a making a list of problems might help?16:34
johnthetubaguysee which ones are the same thing, which a very different?16:35
cdentIt’s not just “how do we make sigs successful”16:35
johnthetubaguyin my head, its how do we get the community onto a more sustainable footing16:35
johnthetubaguybut that doesn't cover all the aspects here16:36
sdagueso... given infinite time, it seems like the thing to do would be actually interview both current and former openstack devs and ask them about what they perceive as their top challenges for themselves, and for others to be successful here. And basically build a research report on that. The SDK one was pretty enlightening, and it was done from a perspective of lets just go research.16:39
sdaguehow to carve out time for such a thing... *punt*, but at least throwing that out there16:40
cdentwhat you both just said there may provide a reasonable fulcrum to get started:16:40
* johnthetubaguy nods +1 but runs away to cook his dinner16:42
cdentif we agree that regular devs (and other members of the community) experience top challenges which make the sustainability of the community difficult _and_ we aren’t clear on what they are then we need to be able to make some time to do the research16:42
cdentthe research being: interviews or something of that ilk16:42
sdaguecdent: yeh16:42
cdentso we can say to the board (and ourselves and others) “we need to figure out how to do this thing”16:43
cdentthat’s at least a starting point16:43
sdaguecollette did a similar thing early on which resulted in "we need to do a vision"16:43
cdentyeah, and though that vision isn’t perfect it has resulted in some clearly useful initiatives that have already had some positive results16:43
fungiwe need to not delude ourselves into assuming the "top" reported problems are necessarily the ones which need to be (or even possible to be) fixed though16:44
cdentI think that’s why this needs to be formulated as interviews and not surveys16:45
fungilots of people when asked for feedback on challenges they encounter will (often just subconsciously) have started trying to come up with their own solutions and will report the lack of support for their particular solution rather than the actual problem which sent them down that path16:46
cdentit is useful and important to be unabashedly biased in who and what gets interviewed and how the analysis is done16:46
fungiit's also very easy to ask the wrong questions16:47
* cdent must dash to cut grass before rain16:48
fungii'm reminded of when the diversity working group, instead of interviewing/surveying people to find out what prevented actual diversity in our community, rather asked what made it hard for them to get involved. the answers were all the common ones you would expect from people who had been assigned by their employers to participate in a free software project for the first time16:49
fungiso none of the insights from that research were in any way helpful for improving diversity and were instead about figuring out how to make free software contribution easier for newcomers in general16:51
fungimost of which tended toward making it more like the commercial/proprietary software development they were already used to16:52
sdaguefungi: I agree with all of that, but it is the point cdent makes. You do this as interviews (which ideally you record the audio of so you can listen back later and digest)16:52
sdagueand the analysis of the levers that exist there are part of the human intellegence of it. Because people don't always know what their actual issues are, but if you ask about positive and negative experiences and to describe what happened, you could get to the heart of things sideways16:53
fungitotally. the anecdote was meant to reinforce his assertion16:54
*** alex_xu has quit IRC17:05
dhellmannit looks like I missed an interesting office hour today :-/17:08
*** alex_xu has joined #openstack-tc17:13
*** cdent_ has joined #openstack-tc17:56
*** cdent has quit IRC17:59
*** cdent_ is now known as cdent17:59
*** Rockyg has quit IRC18:12
*** cdent has quit IRC18:55
fungidhellmann: the "wednesday" one was pretty good too19:20
dhellmannfungi : yeah, I need to make a point of clearing my schedule at these times. Today I was working a bug, so it couldn't be helped.19:21
fungithe oslo.config one?19:21
fungibut yeah, i put all three office hour times on my reminder list so even if i can't make it i know it's going on19:22
dhellmannfungi : yeah19:25
*** cdent has joined #openstack-tc19:41
*** openstack has joined #openstack-tc19:59
*** cdent has quit IRC20:06
*** openstackgerrit has quit IRC20:17
*** marst has quit IRC22:19

Generated by 2.15.3 by Marius Gedminas - find it at!