Tuesday, 2017-08-22

*** emagana has quit IRC00:21
*** gcb has joined #openstack-tc01:36
*** gcb has quit IRC02:12
*** gcb has joined #openstack-tc02:15
*** gcb has quit IRC05:21
*** gcb has joined #openstack-tc05:23
*** jpich has joined #openstack-tc07:34
*** dtantsur|afk is now known as dtantsur08:36
*** cdent has joined #openstack-tc08:36
cdentit’s office hours time09:00
* cdent puts out the sign09:00
cdentAn issue I’d like to get on an agenda somewhere, probably related to the strategic goal of simplification (and perhaps top 5 list and friends): More assertively enabling refactoring and tech debt handing.09:02
cdentThis is a story as old as the world, but it’s commonality makes it no less important.09:03
cdentNew contributors come along to look at OpenStack and in some cases are confronted with code that is so dense, so complex, so hard to change that they choose to go away.09:03
cdentSome of this is the result of organic growth, some of it is the result of less than great architectural decisions, but some of it is simply that we don’t follow simple rules of thumb when writing code:09:04
cdentthings like extracing methods, keeping methods short, avoiding side effects, keeping modules short09:05
gcbcdent: oslo have some deprecated stuff to clean up,  does that  belong to your plan ?09:07
cdentTo some extent, yes. That’s the sort of cleaning up that thingee has been trying to drive with his work on the simplification strategic goal.09:08
johnthetubaguycdent: I like the assumption or architectural decisions09:09
johnthetubaguycdent: I honestly assumed it was the users/operators that need the simplification more, but agreed the dev side is important09:10
cdentPart of this thinking stems from my own exhaustion with just how incredibly hard it is sometimes to get stuff done.09:10
johnthetubaguycertainly much we want to do hits up against tech debt (in all its many forms)09:11
* cdent nods09:11
*** evrardjp has joined #openstack-tc09:12
johnthetubaguyI see placement work and cells work as tech debt removal efforts, I mean they are more than that, but it unblocks so much future work09:12
cdentyes, except we’re constantly uncovering more as we inch forward.09:13
johnthetubaguyI guess my question is how do we sell the value of the work, and how best to keep track of it so those who are interested / value it can easily pick it up09:14
cdentit is good that we are uncovering it (very good), but the list, oh my09:14
cdentwell, that’s part of why I listed the “things like…” above. Those are measurable things which are somewhat agreed upon as ways to inspect code for smelliness09:15
cdentwhen we have limited review bandwidth we often treat those kinds of fixes with disdain, but I think that’s a bad choice09:15
johnthetubaguyI duno, I quite like those little reviews, my problem is finding them in the big haystack of doom with 17 people screaming for attention09:16
johnthetubaguyfor clarity, merging more of that good stuff gets my support, I am just wondering how best to make it happen09:17
cdentwe could expose code metrics a bit more globally09:17
cdentunit and functional coverage results, radon results, that sort of thing09:18
cdentrather than advertising lines of code added, we could trumpet about lines of code removed09:20
johnthetubaguyits worth trying, I don't remember why some of those got pulled down09:29
johnthetubaguywho is doing the lines of code added? I can't remember where those stats are now, which feels bad.09:30
cdentI’m not sure it is a regular thing, but it is something that gets talked about at the end of a cycle “In pike there were N commits from X people, adding M lines of code to the total Y million lines of code that make up openstack” or whatever09:32
cdentbasically I think we need to do a bit of culture jamming09:33
johnthetubaguyI guess if we get out there first with better stats, they get into the marketing09:35
* cdent nods09:37
johnthetubaguycdent: been thinking more about users, working much more with users in the new job, as I hoped.09:51
cdentany surprises so far?09:51
johnthetubaguyapart from seeing "little niggles" turn into massive mountain sized road blocks for new comers I don't have anything that big09:51
johnthetubaguysaw a deploy that moved to keystone v3, but horizon is till only giving our v2 RC files, that was... frustrating09:52
johnthetubaguybut yeah, looking forward to learning a lot more09:53
johnthetubaguycertainly feels like that getting from the point of pressing buttons to getting a good undertanding of how things actually works, and the context around it, so you know what to do09:54
johnthetubaguyseems like a bigger hurdle than you would hope for09:55
* cdent nods09:55
johnthetubaguyoh, I saw someone had deleted an instance created by sahara in an ironic based cloud09:59
johnthetubaguythat left behind the port, and a partial heat stack10:00
johnthetubaguywhich stopped some VMs booting, because the port was trying to bind to the same mac as the existing port10:00
johnthetubaguy...technically correct behaviour in isolation, together, well, <insert you fav nasty word>10:02
johnthetubaguyI think better supporting applications / layering of services is probably the most important missing functionality, like zaneb said I guess10:03
* cdent nods10:04
*** dtantsur is now known as dtantsur|bbl10:40
*** alex_xu has quit IRC11:02
*** alex_xu has joined #openstack-tc11:06
fungicdent: after the leadership summit in boston the bod, tc and uc members present determined that simplification was one of the top 5 topics we needed to address. unfortunately when it came time for volunteers only thingee raised his hand. he's been making some great progress but i expect he'd appreciate assistance11:55
cdentfungi: I’m aware of that. And I think it’s important, and I want to help, but like many people I’m overtasked. Also, while I applaud the efforts to simplify for users/deployers, the wheel I’m squeaking is complexity for contributors11:57
fungiand i do think from a marketing and publicity standpoint we tend to focus too much on what we added in a given cycle rather than on what we removed. maybe that would help get some more people interested or at least looking at things from that angle11:58
cdentWhile I think that user experience always trumps developer experience, the mandate I’ve taken on for myself as a tc person is dev experience (at some point I said something akin to “union rep”)11:58
fungicdent: yeah, on the simplicity for contributors front, i'm glad we were finally able to drop the tie-in to foundation membership11:59
cdentyeah, I think if we can have some kind of blurb (somewhere, i’m not sure where) that is the “here’s what we removed” excitement, that would be fun11:59
cdentyes, that was a _huge_ win11:59
fungii'm setting my sights on the cla to dco move in the next couple cycles as another major hurdle11:59
fungiand working on condensing down to a single single-sign-on ;)12:00
fungiyeah, single-sign-on doesn't help too much if you have more than one12:01
cdentAll that stuff will be a great help, thanks for holding that flag.12:01
fungithingee is collaborating with the foundation's web dev team to revamp the onboarding documentation into more of a prompted webapp which can steer new contributors more effectively to their areas of interest and make sure they set up the tools they need to be prepared for that12:02
cdentThat’s like the front door stuff. Once through the door, it’s a bit of a situation of “wow, how do I get to the kitchen from here"12:02
fungiand the openstack upstream institute effort has been picking up steam and the team on that is trying to get the materials/coursework more reconsumable so that volunteers can easily step up to give the courses at user group meetings or within companies12:05
fungibut i agree, that's all front-door12:05
cdentIt may be that by simply getting more people through the door the available pool of people will include hardy souls who are invested in cleanup. That will get them to the kitchen.12:06
cdentBut then we also need to make sure there is sufficient willingness on the part of reviewers.12:06
cdent(our complexity is complex)12:06
fungithe more people who make it through the front door, the less we'll notice all the ones who get lost on their way to the kitchen, accidentally lock themselves in the basement and starve to death? ;)12:07
fungiyeah, the front-door solutions are (relatively speaking) easier because they apply a lot more globally across the contributor base. most of the intra-project complexity which hampers contribution varies a lot and so is probably harder to address with generalized solutions12:08
* cdent nods12:09
funginot to say that we shouldn't also work to improve there12:09
fungibut i think it needs already embedded contributors to spot those complexities and address them or at least surface them so others know where to focus12:10
cdentsometimes, though, I think people who are embedded get used to it. it becomes normal.12:12
fungiit does require to put on newcomer glasses and look around. i agree that's also not easy12:13
fungiand trying to keep track of what issues new contributors raise and ask for help addressing12:13
smcginniscdent: +1 for tackling technical debt and working on developer simplification.12:54
smcginnisI think developer simplification will naturally lead to some end user simpification.12:55
cdentYeah, I think (or at least hope) so too12:56
*** cdent has quit IRC13:30
*** hongbin has joined #openstack-tc13:59
*** marst has joined #openstack-tc14:11
*** dtantsur|bbl is now known as dtantsur15:13
openstackgerritSean McGinnis proposed openstack/project-team-guide master: Add reno usage for editing stable branch notes  https://review.openstack.org/49631817:07
openstackgerritSean McGinnis proposed openstack/project-team-guide master: Add reno usage for editing stable branch notes  https://review.openstack.org/49631817:10
openstackgerritAndreas Jaeger proposed openstack/governance master: Update docs URLs  https://review.openstack.org/49632117:14
openstackgerritAndreas Jaeger proposed openstack/governance master: oslo-incubator is retired, mark it as such  https://review.openstack.org/49632517:18
*** jpich has quit IRC17:28
*** dtantsur is now known as dtantsur|afk18:05
*** cdent has joined #openstack-tc18:47
cdentweak tc report his week19:09
smcginniscdent: Maybe too many other things going on at this point in the cycle?19:12
cdentyeah, I think that’s probably the case19:12
cdentbut I know myself well enough that if I skip one because there’s not enough going on that week I’ll end up breaking the habit19:13
smcginnisI know what you mean. :)19:17
openstackgerritAndreas Jaeger proposed openstack/governance master: oslo-incubator is retired, mark it as such  https://review.openstack.org/49632519:43
openstackgerritAndreas Jaeger proposed openstack/governance master: Update docs URLs  https://review.openstack.org/49632119:44
openstackgerritSean McGinnis proposed openstack/project-team-guide master: Add reno usage for editing stable branch notes  https://review.openstack.org/49631820:26
openstackgerritChris Dent proposed openstack/governance master: Add Reduce Development Complexity top-5  https://review.openstack.org/49640420:30
cdentjohnthetubaguy, fungi, smcginnis ^20:31
cdentvery WIPpy20:31
cdentsmcginnis: yeah, the packaging of the idea is where I struggled20:35
smcginniscdent: Yeah, I can't think of a good way. Maybe we need to come up with some semi-arbitrary metrics in order to decide when we've "completed" the work.20:38
cdentas if anyone would be able to agree on that? I mean: I would say no modules > 1000 lines as an easy target20:38
cdentwhich means most of nova needs a rewrite20:38
smcginnisOr as an alternative for a top-5 goal, have some way to positively reinforce code cleanup work.20:38
smcginnisTrue. :)20:38
smcginnisMaybe like a stackalytics option that shows a ranking of people that have removed the most lines of code per release.20:39
cdentYeah, I’m not sure if top5 is the right way to go but I figured it was a reasonable way to concretize the discussion20:39
cdentthat’s but one of many metrics, and scarily easy to game to very bad results20:40
smcginniscdent: Yes, definitely good to get the conversation going.20:40
smcginnisRight, any kind of metric we want to guage off of, someone is going to find a way to get their name higher on the list.20:40
*** cdent has quit IRC21:30
*** marst has quit IRC22:22
*** hongbin has quit IRC23:13

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!