Working Group Minutes/EWG 2011-11-21

From OpenStreetMap Foundation

Attendees

IRC nick Real name
apmon Kai Krueger
Firefishy Grant Slater
RichardF Richard Fairhurst
TomH Tom Hughes
zere Matt Amos

Summary

  • Items from F2F board meeting:
    • Clear list of development tasks and motivation (e.g: bounties) for completion.
    • EWG's role in front-page redesign.
    • Identify and fill gaps in downstream data offerings.
  • apmon suggested cleaning up and re-using the wiki top ten tasks page, and volunteered to help with that.
    • "Top ten" tasks were suggested, discussed and voted upon to get the ten roughly considered the "top" ones.
    • Will revisit these and their inclusion in the top ten page on a monthly basis.
  • Some discussion of motivation, including bounties, but time ran out so it was postponed to the next meeting.
  • It was noted that PBF planets have been generated "experimentally" for a while and it probably wasn't really experimental any more. Firefishy removed the experimental label.

IRC Log

18:00 < zere> minutes of the last meeting, for your reading pleasure: http://www.osmfoundation.org/wiki/Working_Group_Minutes/EWG_2011-11-14
18:00 < zere> any objections, please let me know.
18:00 < TomH> evening
18:02 < zere> agenda for today: last meeting actions, stuff from the board f2f, AoB.
18:02 < zere> the f2f minutes still haven't been published, but it can't wait another week.
18:03 < zere> looks like the only action from last week's minutes is on me and RichardF to test apmon's packages.
18:04 < zere> i tried them this morning, but unfortunately the rails3 branch had gone away by that point, so the package didn't 'configure' properly. apmon has updated the package, but it'll take a while to go through launchpad's build system.
18:04 < apmon> morning
18:06 < zere> so, on to items from the board meeting.
18:07 < zere> the board have asked us to look at making a clear list of development tasks and consider the possibility of offering bounties for them.
18:07 < zere> also, to see if there's a gap in the offerings by geofabrik et al, in terms of useful exports for downstream users.
18:08 < TomH> yeah because what we really need it lots of GSOC wannabees writing crap code for cash
18:09 < zere> there's going to be a front page redesign, and we have a role to play in making sure that the redesign gets merged in in a sympathetic way and doesn't break a bunch of stuff.
18:09 < apmon> Having a clear list of development tasks sounds like a good idea
18:09 < apmon> That allows people to have a look at the list and see if they can help with any of them
18:09 < zere> yes. and as a group i'd like to see if we can't get a set of those out this meeting on the wiki.
18:11 < apmon> zere: E.g. clean up http://wiki.openstreetmap.org/wiki/Top_Ten_Tasks ?
18:11 < zere> as to the crap code aspect, i think we have to review any submission for any bounty and, if it's lacking in quality, help explain what needs to be done to fix it.
18:11 -!- Firefishy [~osm@94-194-60-19.zone8.bethere.co.uk] has joined #osm-ewg
18:11 < zere> apmon: yes.
18:11 < Firefishy> greetings.
18:12 < TomH> zere: well sure, but those reviews are very time consuming and we have a bandwidth problem there
18:12 < zere> greetings, Firefishy. our topic of discussion is to find the "top ten tasks" that we think need doing.
18:12 < zere> for some values of "ten".
18:12 < apmon> zere: Sounds good to me.
18:12 < RichardF> sorry, bit late. Was away at the weekend and didn't get chance to test apmon's package (as it were) - but am sure zere has managed it with aplomb
18:13 < TomH> plus my concern is that we will get lots of students with no experience of our platform who are only in it for the money
18:13 < zere> i have a similar worry.
18:14 < zere> the question is; is it worth offering *something* for completing the tasks. if so, is that thing money or something else?
18:14 < zere> personally, i think money isn't the right kind of motivator.
18:15 < apmon> Do you have other ideas?
18:15 < RichardF> I'd agree, but I can see the argument that it can work where you have a certain very specific short-term, well-specified goal.
18:15 < zere> (i should point out we have a $1000 budget at this point - we have discretion over how it's spent: bounties, prizes, leaflets, etc...)
18:15 < RichardF> I suspect any bounties would have to be pretty substantial - J Random Peruvian is unlikely to be able to get up and running with OSM in a short time if the questions on help &c. are anything to go by.
18:16 < apmon> It is probably worth figuring out the tasks first. Then depending on what those are, different types of motivation might be appropriate
18:16 < RichardF> but really, bounties need only be a last resort if our existing community won't fix these putative top ten tasks
18:16 < RichardF> apmon: exactly
18:16 < zere> i don't know if this would work, but i'd like to try prizes in the form of a t-shirt / goodie bag, plus entry into a leaderboard and the possibility of interview / profile with CWG.
18:16 < zere> ok. let's come back to the motivation question.
18:17 < zere> any ideas for the top ten tasks? just shout 'em out and we'll come back at the end to discuss them.
18:18 < RichardF> finishing OSB(-like) integration
18:18 < zere> routing (frontend and backend), ODbL migration tools
18:19 < RichardF> cgimap for writes?
18:19 < RichardF> (i.e. changeset uploads)
18:19 < zere> deleted items API call
18:20 < zere> (XML API, before RichardF claims that one)
18:20  * RichardF laughs
18:20  * RichardF resists the temptation to say "finish Jangaroo/Falcon-JS so we can deploy P2 in JS"
18:20 < apmon> Better handling of overlapping areas in the data browser
18:21 < RichardF> dynamic clickable POIs, Google-style, using Mapnik metawriter
18:21 < TomH> apmon: that's an OL issue, not an OSM issue
18:21 < apmon> i18n in P2 and help
18:21 < zere> i don't think we restrict ourselves to OSM-only issues at this point.
18:21 < RichardF> move the Mapnik stylesheet to something sane (Carto or whatever) so patches are more feasible
18:22 < RichardF> slight curveball: awesome OSM editor for iPhone and Android
18:23 < apmon> Better integration with the QA tools
18:23 < RichardF> OWL-powered History^H^H^H^H^H^H^H Activity tab
18:24 < RichardF> (and API calls to support it so P2 et al can benefit from the same stuff)
18:24 < RichardF>  /way/full calls at a particular timestamp (like P1)
18:27 < zere> suppress inactive users in 'users near me' view?
18:29 < zere> ok. have we run out of steam?
18:30 < apmon> Are there anymore smaller usability improvements people know of that could be added to this list?
18:32 < zere> i see: 1) osb/notes integration, 2) routing (backend), 3) routing (frontend), 4) ODbL migration tools, 5) cgimap support for uploads, 6) XML deleted items call, 7) OL "fix" for overlapping areas, 8) clickable POIs on mapnik map, 9) i18n in PL2, 10) i18n in help, 11) convert mapnik style to carto, 12) mobile OSM editor, 13) integration with QA tools, 14) OWL-powered history tab, 15) way/full calls at a point in time.
18:33 < apmon> Ordering being somewhat with respect to importance?
18:33 < zere> apmon: i dunno. are you aware of any?
18:33 < zere> (UX improvements, not ordering)
18:33 < apmon> I can't think of any for now, but I am sure there are some
18:34 < zere> me too.
18:34 < zere> well, we can always come back every few meetings to review this list.
18:34 < apmon> But I am sure once we announce the list to try and get people to consider implementing them, loads more suggestions will be thrown our way
18:35 < zere> so the list above was in order that it was suggested according to my IRC display. i reckon we discuss each and then figure out "importance" later.
18:35 < zere> apmon: yes.
18:35 < RichardF> there's a million and one things for P2 - I wasn't listing them but trac is a good start...
18:37 < apmon> RichardF: Are there any major ones? Might be worth listing a couple here too
18:38 < zere> oh, there's many "major" ones... i dunno if any are really major, though.
18:38 < RichardF> apmon: tutorial mode, better help, load from local disk, better indexing to cope with large datasets, and a zillion UI improvements, starting with the drag-and-drop POI panel
18:39 < RichardF> but, in general, I'm reasonably happy with how P2 development is progressing - I suspect the osm.org site itself is more of a priority
18:39 < apmon> Not sure if it fits here, but a non-experimental pbf export of the planet-file might be good
18:39 < zere> ok. bearing in mind the top 10, what do we think about osb/notes integration?
18:40 < zere> apmon: it's been experimental for a while now... probably just needs renaming. Firefishy will know best.
18:41 < RichardF> zere: hugely significant, I think
18:41 < Firefishy> apmon: If others agree, I'll rename the pbf stuff.
18:42 < Firefishy> But will still going to run secondary to the xml planet.
18:42 < zere> that's good. i was about to say that no-one thought it important enough to say anything about :-)
18:42 < zere> Firefishy: sounds good to me. i haven't heard any complaints about pbf being broken.
18:42 < apmon> zere: TomH will know best about the status, but I think it might be reasonably close to being ready 
18:44 < TomH> what?
18:44 < apmon> osb/notes integration
18:45 < TomH> err yeah
18:45 < TomH> not exactly
18:45 < TomH> server code is mostly sorted but client side needs a rewrite
18:46 < zere> ok. so there's enough work to put it on the top 10.
18:46 < TomH> but I decided it would be easier to do against rails3
18:46 < apmon> TomH: Well, let me know if there is anything I can do with the front-end rewrite
18:47 < zere> ok. routing frontend: worth doing without having sorted the back-end?
18:47 < Firefishy> planet.osm.org: pbf-experimental renamed to pbf and hidden redirect enabled.
18:48 < RichardF> is there a plan for sorting a routing backend?
18:48 < zere> not exactly. dennis is coming (hopefully) to the hack weekend and i intend to pick his brains
18:48 < zere> about how practical it would be to have multiple routing profiles updated at least daily, etc...
18:48 < apmon> I'll probably get back to the routing frontend stuff soonish
18:49 < apmon> It was one of the things I was considering to work on during the hack-weekend 
18:50 < apmon> There is still the option of gosmore for routing profiles like pedestrian and biking
18:50 < apmon> It's main disadvantage of not scalling to well, shouldn't matter much for the shorter distances of walking / biking
18:51 < RichardF> hey, we should be relegating cars to the secondary engine, put cycling on the good engine ;)
18:51 < Firefishy> OSRM is getting configurable profile support. See: https://github.com/DennisOSRM/Project-OSRM/blob/master/speedprofile.ini
18:53 < apmon> RichardF: gosmore has sofar been more feature rich than OSRM, just not as fast for long distances
18:54 < RichardF> cool :)
18:54 < apmon> OK, what about the other points?
18:55 < apmon> adding write capabilities to cgi-map is presumably for performance issues
18:56 < RichardF> yep. (though to be fair the situation seems improved at the moment.) I suspect it may be Very Hard but zere would know more.
18:56 < apmon> The backend servers don't seem to particularly loaded, so it would mostly be a question of perceived  upload latency to users
18:57 < apmon> Bulk uploading sounds like it takes quite a while, but is the latency an issue currently for more normal users?
18:59 < apmon> Which of the other points listed are considered high priority?
18:59 < RichardF> I used to have fairly serious problems with changeset uploads taking so long that the response got lost on the return journey to P2, usually when large-ish relations were involved. but, like I say, it seems better now.
19:00 < zere> for fun, here's a poll: https://docs.google.com/spreadsheet/viewform?formkey=dDhpd3BYdVRRRmxhdjBIOWFOWFoxUnc6MQ
19:00 < zere> each of the items from 1 to 5 (least to most) important. just a straw poll.
19:01 < zere> and yes, adding anything needing auth to cgimap isn't trivial... suspect it might need major backend changes to handle that.
19:02 < zere> oh, but that would allow us to do the "if you're logged in you don't get rate limited" rule, which is currently pretty impossible to do properly.
19:05 < apmon> zere: not exactly that, but could one "simply" reset the memcahced download counter from the rails_port everytime an upload occurs?
19:05 < RichardF> done :)
19:06 < zere> apmon: doesn't help when someone comes to edit a huge river and hasn't done any uploads yet...
19:07 < zere> i mean, cgimap could look for an oauth header and just assume it was valid...
19:07 < zere> but how long before some scraper figures that one out?
19:10 < zere> has everyone done the poll that wants to? apmon, TomH, Firefishy?
19:10 < apmon> yes
19:15 < zere> the top 10 are OSB/notes, PL2 tutorial mode, PL2 i18n, PL2 help, ODbL migration tools, clickable POIs on frontpage, routing (frontend), XML deleted items call, OWL-powered history tab, and routing (backend)
19:16 < zere> anyone want to take a stab at re-making the "top ten tasks" wiki page?
19:17 < apmon> That shouldn't be too difficult, so if no-one else is going to do it, I can
19:17 < zere> ah, that's what i like. a volunteer :-)
19:18 < zere> it's getting late, so i suggest we leave the discussion of what prize / bounty / accolade is given to the next meeting.
19:18 < RichardF> good plan
19:20 < zere> ok. thanks to everyone for coming. see you in #osm-dev or IRL over the hack weekend!