Humanitarian OSM Team/Working groups/Technical/meeting 2015-03-02 TM2

From OpenStreetMap Wiki
Jump to: navigation, search

Mar 02 17:11:38 mkl: AndrewBuck, pgiraud, russdeffner my connection is too low to connect to mumble
Mar 02 17:12:03 mkl: Was just writing on the chat, we have a lot of good stuff in place ... solid code base, lots of good ideas to implement
Mar 02 17:12:12 mkl: main constraint is time
Mar 02 17:12:41 mkl: What I wanted to suggest, to help keep things moving on project management, deployment, etc, was to institute a regular development cycle
Mar 02 17:12:58 mkl: Say monthly development and deploy to production, on a schedule
Mar 02 17:13:22 BlakeGirardot: I think that is a good idea.
Mar 02 17:13:22 pierregiraud: I said that I often don’t even take time to answer to requests on github or elsewhere
Mar 02 17:13:52 BlakeGirardot: But it still runs into a time issue for reviewing the pull requests
Mar 02 17:14:03 BlakeGirardot: merging and then deployment adn then testing
Mar 02 17:14:16 pierregiraud: for your information, Nitika’s work is not going as well as it should
Mar 02 17:14:37 pierregiraud: we probably will be far away from what was expected during the intership
Mar 02 17:14:39 mkl: part of a monthly schedule could include github issue triage. we had that going for a while
Mar 02 17:15:05 pierregiraud: issue triage hasn’t been a big success actually
Mar 02 17:15:08 mkl: pierregiraud: how much time does management pull requests and deployment take?
Mar 02 17:15:23 pierregiraud: I can’t really say
Mar 02 17:15:25 mkl: in what way pierregiraud? we also need to close issues too :)
Mar 02 17:15:49 BlakeGirardot: To be done right it takes a fair amount of time.
Mar 02 17:16:04 pierregiraud: it does
Mar 02 17:16:05 BlakeGirardot: pierregiraud Keeps the code base very solid and clean
Mar 02 17:16:14 BlakeGirardot: and I trust his merges
Mar 02 17:16:25 BlakeGirardot: and we have had zero issues when he is responsible for them.
Mar 02 17:16:30 BlakeGirardot: I mean code issues
Mar 02 17:16:42 pierregiraud: but we’re currently in a phase where people have been requesting for new features
Mar 02 17:17:01 pierregiraud: that a lot of people agree to say they are required
Mar 02 17:17:14 pierregiraud: though they are not implemented
Mar 02 17:17:23 mkl: well that needs to be translated into a development plan
Mar 02 17:17:40 mkl: sounds like there are a few "resources" needed
Mar 02 17:17:51 mkl: one is to manage issues and build development plans / sprints
Mar 02 17:17:59 mkl: another is to manage pull requests and deploys
Mar 02 17:18:05 mkl: and another of course is to implement
Mar 02 17:18:37 BlakeGirardot: Yes, good point mkl
Mar 02 17:18:39 mkl: all of this is now done ad-hoc
Mar 02 17:20:34 mkl: do we all agree, this is what's needed?
Mar 02 17:20:47 mkl: it's a seperate question on how to make it happen
Mar 02 17:20:52 BlakeGirardot: So
Mar 02 17:20:56 pierregiraud: I agree
Mar 02 17:21:02 BlakeGirardot: let me make sure I understand quickly
Mar 02 17:21:14 BlakeGirardot: 1. manage issues and help with priortization
Mar 02 17:21:25 BlakeGirardot: 2. deployment
Mar 02 17:21:34 sev_hotosm: how many people actually participate to the code?
Mar 02 17:21:35 BlakeGirardot: 3 implementation
Mar 02 17:21:55 BlakeGirardot: Is that what you are saying as seperate items mkl?
Mar 02 17:22:11 pierregiraud: sev_hotosm: I’ve been writting 95% of the code
Mar 02 17:22:28 pierregiraud: more recently a couple of devs have been helping though
Mar 02 17:22:33 BlakeGirardot: and really I guess the order is prioritizaion, implementation, deploy/test
Mar 02 17:23:02 mkl: i've contributed a little lately
Mar 02 17:23:02 sev_hotosm: OK so as of today monthly scheduling would basically depend on your own schedule :)
Mar 02 17:23:15 pierregiraud: sev_hotosm:
Mar 02 17:23:23 pierzen|2: hi all
Mar 02 17:23:36 pierregiraud: tasking manager unfortunately depends on the time I’m able to spend on it
Mar 02 17:23:46 mkl: i think that if we had a monthly schedule for planning and deployment, then it would attract more developers
Mar 02 17:23:55 sev_hotosm: JOSM can afford monthly releases because I guess it has more contributors
Mar 02 17:24:00 sev_hotosm: ah OK
Mar 02 17:24:30 mkl: it's sort of chicken and egg
Mar 02 17:24:45 mkl: development of v2 got us in a good point regarding the code base. it's solid
Mar 02 17:24:57 pierzen|2: We discussed about dev phase first to validate at least for content related developments.
Mar 02 17:25:10 sev_hotosm: except you need more than one chicken to get eggs :)
Mar 02 17:25:59 mkl: maybe this is an English phrase? :) it just means, we are a bit stuck about what comes first
Mar 02 17:26:54 pierzen|2: With the ticket, somebody picks a ticket and do it. But no priority among all the tickets.
Mar 02 17:27:06 mkl: right
Mar 02 17:27:23 mkl: we need to spend time not only prioritizing and choosing
Mar 02 17:27:31 althio[mobile-]: Hi all, just here following on mobile but cannot jointe mumble
Mar 02 17:27:47 mkl: but then provide a "space" for devs to talk to each other about what they work on
Mar 02 17:28:03 mkl: and someone to manage the pull requests and deployments
Mar 02 17:28:20 mkl: for instance, i have done some work on the API lately. but i don't think there's been time to deploy
Mar 02 17:28:48 pierregiraud: The deploy process started last week
Mar 02 17:28:59 pierzen|2: the tech list is good for more regular interaction. I think that this can be the space to discuss.
Mar 02 17:29:04 pierregiraud: I actually prepare a release once in a while when I think it’s work
Mar 02 17:29:06 pierregiraud: no real paln
Mar 02 17:29:08 pierregiraud: plan
Mar 02 17:29:30 pierregiraud: once again, I do it when I find
Mar 02 17:29:31 BlakeGirardot: Ya, for some of these big things we need a staging server for testing/evaluation
Mar 02 17:30:08 BlakeGirardot: I can help with that, I have to problem deploying to a testing server for people to give feedback on. Like the front page redesign and the back end changes that go with it.
Mar 02 17:30:25 mkl: Well one way of doing that is to have a regular time monthly where deployment is considered.
Mar 02 17:30:36 mkl: If there's nothing to release, fine
Mar 02 17:30:43 mkl: But if there is, that's when it happens
Mar 02 17:30:54 mkl: That kind of predicability is helpful for developers
Mar 02 17:32:03 mkl: I think keeping a deployment server is great, BlakeGriardot. Did Drazen set up a server for this?
Mar 02 17:32:35 sev_hotosm: I would suggest in the homepage a space listing the last improvements, like JOSM does. It shows it is a living project
Mar 02 17:32:52 pierregiraud: I ‘m not sure if a testing server will really help with the fact that implementation time is lacking
Mar 02 17:33:08 mkl: pierregiraud: i've been implementing!
Mar 02 17:33:10 BlakeGirardot: No, it won't pierregiraud but it will help somewhat.
Mar 02 17:33:28 BlakeGirardot: It might also provide the space to take on bigger
Mar 02 17:33:30 mkl: sev_hotosm: good idea. though i think on the about page is better for that level of detail
Mar 02 17:33:31 BlakeGirardot: projects
Mar 02 17:33:46 pierregiraud: what about release notes?
Mar 02 17:33:48 BlakeGirardot: like all the planning and ideas around the front page redesign.
Mar 02 17:34:06 pierregiraud: I know the release notes have been poorly filled recently but…
Mar 02 17:34:18 pierregiraud:
Mar 02 17:34:22 sev_hotosm: mkl I would never go to the About page to see the last improvements
Mar 02 17:34:25 pierregiraud: I know it’s not perfect
Mar 02 17:34:39 pierregiraud: where would you go sev_hotosm ?
Mar 02 17:35:22 mkl: I could imagine a link on the front page that just said the last time there was an update, with a link to more detail
Mar 02 17:35:24 sev_hotosm: and basically it would be a satisfaction eg if your contribution was listed mkl
Mar 02 17:35:51 * pierregiraud thinks that what the tasking manager is missing is me being a better project leader Mar 02 17:35:54 sev_hotosm: I would not go anywhere, this is why it is good JOSM put this in the backgroup at start
Mar 02 17:36:18 sev_hotosm: and it would be a satisfaction for other developers
Mar 02 17:36:30 BlakeGirardot: pierregiraud: I disagree 100%. From my perspective you are missing more hours in the day. And I would suggest funding some of your time to work on it.
Mar 02 17:37:08 pierregiraud: FYI I already get paid for developing some of the features of the tasking manager
Mar 02 17:37:29 BlakeGirardot: It still appears to me to be a time issue.
Mar 02 17:37:40 pierregiraud: but I cannot be paid full time when others like you or Pierre B. are spending so much time for HOT as volunteers
Mar 02 17:38:10 mkl: BlakeGirardot: +1
Mar 02 17:38:21 BlakeGirardot: I don't know. Some things are really core to our mission and TM2 is one of those cores.
Mar 02 17:38:30 mkl: I guess, there are a few ways to unblock this
Mar 02 17:39:07 pierregiraud: I wished other devs would know the code and the philosophy behind as much as I do
Mar 02 17:39:31 mkl: pierregiraud: could we schedule a webinar type thing where you could share this?
Mar 02 17:39:51 althio[mobile-]: Well lucky - or unlucky you - this is the bottleneck apparently ;)
Mar 02 17:39:56 BlakeGirardot: How much of it is knowing pyramid pierregiraud ?
Mar 02 17:40:04 pierregiraud: I don’t really know how webinars work
Mar 02 17:40:14 BlakeGirardot: I mean if you learn pyramid does that make a lot more of TM2 code base "easy" to work on ?
Mar 02 17:40:50 pierregiraud: you have to know pyramid, but bootstrap as well
Mar 02 17:40:58 mkl: i mean, set up a time on Google Hangout, for Developers to get a tour of the code with you
Mar 02 17:40:59 pierregiraud: the javascript is currently very messy
Mar 02 17:41:03 BlakeGirardot: pyramid, sqlalchemy, mako, bootstrap. Ok. Interesting.
Mar 02 17:41:26 pierregiraud: BlakeGirardot: you got it
Mar 02 17:41:36 althio[mobile-]: I proposed the staging server as a place to educate others for dev and deploy
Mar 02 17:41:41 pierregiraud: this is not that difficult for python web developpers
Mar 02 17:42:32 pierregiraud: althio[mobile-]: if you get something working on your machine, it doesn’t really have to be seen live
Mar 02 17:42:47 BlakeGirardot: I think mkl has a pretty good idea there though. A walk through of the code, even though some of us (me) are not python pyramid devs would help.
Mar 02 17:42:58 pierregiraud: the problem with a dev server is how people can get their code running there
Mar 02 17:43:14 mkl: I think BlakeGirardot was offering to manage staging server
Mar 02 17:43:28 pierregiraud: with ssh access to anyone?
Mar 02 17:43:35 BlakeGirardot: Ya, I am. The staging server really help for larger/longer term where you need others
Mar 02 17:43:42 BlakeGirardot: to see what you ahve done and provide feedback
Mar 02 17:44:11 pierregiraud: would it be a sandbox?
Mar 02 17:44:33 BlakeGirardot: I will take your and dodobas' advice on how best to handle that pierregiraud
Mar 02 17:45:03 pierregiraud: I don’t think dodobas will advice to let people connect to the given server
Mar 02 17:45:19 pierregiraud: I don’t even have access to the production nor testing server myself
Mar 02 17:45:20 BlakeGirardot: as a branch from tm2 or a fork that keeps master synced? I dont know, but I can help with it it if will help with our dev cycle.
Mar 02 17:45:30 BlakeGirardot: I don't want anything to do with production :)
Mar 02 17:45:47 pierzen|2: Correct me on this, but Dev server is not that expensive. You have host services at less then 100$ a year.
Mar 02 17:46:00 pierzen|2: We should find budget for this.
Mar 02 17:46:17 BlakeGirardot: I think that is probably about right pierzen|2 maybe $150 a year
Mar 02 17:46:21 pierregiraud: pierzen|2: once we have a technical way to handle it
Mar 02 17:46:45 BlakeGirardot: It is really the repository that is the sticking point on the dev/staging server.
Mar 02 17:46:50 pierregiraud: I think this is a good option tough
Mar 02 17:47:18 pierregiraud: though
Mar 02 17:47:24 mkl: i think we already have servers which could handle this. it's just a matter of setting up a vhost
Mar 02 17:47:32 BlakeGirardot: api dev is another example where a dev/staging server helps.
Mar 02 17:50:54 mkl: pierregiraud: what do you think about a regular plan/deploy cycle?
Mar 02 17:51:08 pierregiraud: +1
Mar 02 17:51:13 mkl: if there's nothing to deploy, then np
Mar 02 17:51:31 pierregiraud: I’ve tried in the past to release the application every 1.5 month or 2
Mar 02 17:51:40 pierregiraud: it can be more strict
Mar 02 17:52:51 pierregiraud: realeasing more often is better I think
Mar 02 17:53:05 mkl: how much time do you have for this, if any?
Mar 02 17:53:10 pierregiraud: if I get due dates for releases I’ll probably do a bettre work
Mar 02 17:53:23 pierregiraud: releasing doesn’t take much time actually
Mar 02 17:53:35 pierregiraud: let’s say 2 hours or so
Mar 02 17:53:39 mkl: for me, it also means i can schedule in specific time to help
Mar 02 17:53:46 pierregiraud: dodobas helps also for the releae
Mar 02 17:53:54 pierregiraud: he runs the translation process
Mar 02 17:54:17 pierregiraud: and deploys on the server
Mar 02 17:54:44 pierregiraud: should I also review the tickets or pull request on regular basis?
Mar 02 17:54:58 mkl: i'd be willing to help with this too
Mar 02 17:55:32 mkl: there are A LOT of tickets :)
Mar 02 17:55:38 pierregiraud: indeed
Mar 02 17:56:31 pierzen|2: We also need to agree on the Redisign in particular of the main page and other priorities.
Mar 02 17:57:11 mkl: I think there's a good agreement on front page, though it might not be organized in one requirements spec
Mar 02 17:57:20 mkl: There the blocker is definitely implementation time
Mar 02 17:57:30 mkl: Sounds like the mentorship didn't work out as planned
Mar 02 17:57:40 pierregiraud: not it didn't
Mar 02 17:57:59 mkl: I know I wouldn't mind picking off some front page stuff, as I also work on the API
Mar 02 17:58:13 mkl: we just need to organize
Mar 02 17:58:27 BlakeGirardot: And some staging so people can give feedback on it as the dev goes along
Mar 02 17:58:27 pierregiraud: let’s wait until the internship is finished
Mar 02 17:58:40 BlakeGirardot: When does that end pierregiraud ? I thought Thursday?
Mar 02 17:58:54 pierregiraud: 9th of march
Mar 02 17:59:07 BlakeGirardot: Ah, ok, I thought 4th. Cool.
Mar 02 17:59:27 pierregiraud: but I don’t expect anything at all for the next week
Mar 02 17:59:31 pierzen|2: To monitor Tasks outside of TM, either access the DB directly or query through an API. Would it be a problem to access the Db directly?
Mar 02 18:00:03 pierregiraud: I will never allow direct access to the DB directly
Mar 02 18:00:06 mkl: This is what the API is for. Direct DB access isn't good
Mar 02 18:00:07 BlakeGirardot: pierzen|2: That is part of what mkl is working on, and is actually on the dev server now. .json files of project/projects/tasks
Mar 02 18:00:28 mkl: I have to run to lunch. Sounds like we've made a lilttle progress identifiying actions to get wheels turning
Mar 02 18:00:29 althio[mobile]: I don't know where you stand about triage and priorities... Anyway I gave my input, hopefully you can use it as a start for priorities discussion or triage session
Mar 02 18:00:54 pierzen|2: direct db read only, is this even a problem?
Mar 02 18:01:10 pierregiraud: yes
Mar 02 18:01:19 pierregiraud: some projects are private
Mar 02 18:01:33 mkl: and then people can run strange queries
Mar 02 18:01:48 pierregiraud: which can be expensive
Mar 02 18:01:54 pierzen|2: or ...
Mar 02 18:01:56 BlakeGirardot: You will be able to get all the current projects, their status and the individual task info as well via a GeoJson file, which is then easy to work with.
Mar 02 18:02:01 mkl: this is the point of the API
Mar 02 18:02:03 pierregiraud: we don’t want the tasking manager to run out of load
Mar 02 18:02:23 mkl: API is getting pretty well documented, in the OSMTM wiki and in a couple issues
Mar 02 18:02:43 mkl: so it's just me, or someone, finding time to implement
Mar 02 18:02:57 mkl: it's a priority for me to get the basics for projects done in the next little while
Mar 02 18:03:00 mkl: this month
Mar 02 18:03:01 pierregiraud: oh I remember this documentation
Mar 02 18:03:31 pierregiraud: the wiki will have to be updated when it’s in production
Mar 02 18:03:36 mkl:
Mar 02 18:03:45 BlakeGirardot: I have to get with you about that mkl, I think there are a couple of issues with what is checked in so far. And it is on the dev server if you want to look.
Mar 02 18:04:00 mkl:
Mar 02 18:04:11 mkl: ok, let's touch base about this soon
Mar 02 18:04:14 mkl: i gotta run
Mar 02 18:04:19 pierregiraud: i gotta run too
Mar 02 18:04:22 BlakeGirardot: tyvm mkl
Mar 02 18:04:30 BlakeGirardot: thank you all very much for the meeting time.
Mar 02 18:04:31 mkl: let's schedule next triage session and deployment
Mar 02 18:04:38 mkl: over email i guess
Mar 02 18:04:47 mkl: thank you BlakeGirardot for organinzing!
Mar 02 18:05:04 pierregiraud: yes thanks
Mar 02 18:05:35 pierregiraud: bye all
Mar 02 18:05:47 BlakeGirardot: Bye, cheers, see you on the tech wg email list :)
Mar 02 18:06:38 pierzen|2: bye all
Mar 02 18:08:27 dodobas: err, mail said 18:00 utc
Mar 02 18:09:22 dodobas: ahh.. crap... my bad
Mar 02 18:09:29 althio[mobile]: Hi dodobas ;)
Mar 02 18:09:43 dodobas: crap crap
Mar 02 18:10:18 dodobas: hi althio[mobile]
Mar 02 18:10:28 BlakeGirardot: dodobas: I sent out the wrong time?
Mar 02 18:10:43 sev_hotosm: pierregiraud the TM development is still funded?
Mar 02 18:10:51 dodobas: no.. it was correct for the training ...
Mar 02 18:11:09 dodobas: i just 'read' it as twg... blah
Mar 02 18:11:32 dodobas: will read the backlog and comment on twg ml
Mar 02 18:11:43 dodobas: my bad
Mar 02 18:11:46 BlakeGirardot: I see. I did also send one out with the correct time for this on Feb 22 :)
Mar 02 18:11:47 althio[mobile]: dodobas: as they say... Bye, cheers, see you on the tech wg email list :)
Mar 02 18:12:05 BlakeGirardot: No big deal, email works too.
Mar 02 18:12:27 BlakeGirardot: Sorry for the confusion
Mar 02 18:12:56 mataharimhairi: Very informative! Good to see where everyone stands with the Tasking Manager :)
Mar 02 18:13:02 mataharimhairi: I feel like a fly on the wall.
Mar 02 18:13:04 mataharimhairi: Ciao
Mar 02 18:13:27 althio[mobile]: Ciao