Working Group Minutes/EWG 2014-11-10

From OpenStreetMap Foundation

Attendees

IRC nick Real name
apmon Kai Krueger
pnorman Paul Norman
RichardF Richard Fairhurst
zere Matt Amos

Summary

  • meeting time
    • zere had (incorrectly) started the meeting at 18:30 UTC, but the meeting is supposed to start at 17:30 UTC.
  • osm2pgsql
    • pnorman has been investigating issue 208 and has found that ORDER BY is substantially faster than CLUSTER.
    • this might make implementing partitioning more difficult, but it isn't insurmountable.
  • openstreetmap-website
    • zere has fixed some HTML & visual issues with the routing branch. now waiting on review.
    • the overpass query branch was merged, and pushed live.
    • ACTION: zere to find out if it was a soft-launch or whether it can be publicised.

IRC Log

18:35:29 <zere> apologies, minutes of the last meeting aren't written up yet - not that there's much to write up... the raw log is here: http://matt.dev.openstreetmap.org/osm-ewg/2014/osm-ewg.2014-11-03-18.33.log.html
18:35:51 <pnorman> The wiki page says 17:30 UTC
18:37:15 <zere> i haven't received anything for today's agenda. other than that i hate web browsers and CSS. i'm not sure how we got into this mess...
18:37:37 <zere> pnorman: you're right! oops... okay, so the meeting *should* have been an hour ago
18:38:06 <zere> #topic AoB
18:38:11 <pnorman> I looked at https://github.com/openstreetmap/osm2pgsql/issues/208, and ORDER BY is substantially faster than CLUSTER.
18:38:59 <pnorman> My understanding is ORDER BY poses some issues if we do partitioning in the future, but I don't know much about that
18:40:20 <zere> okay, cool. my guess is that only a small number of people will want to do partitioning anyway?
18:40:38 <apmon> wouldn't partitioning be universal?
18:40:56 <apmon> just like we already "partition" the lines table into a roads table?
18:41:56 <apmon> It might be that the partitioning becomes customizable to tune it to the style sheet, but I would assume that (nearly) everyone would want some form of partitioning.
18:43:40 <apmon> If I remember correctly, the issue with ORDER BY and partitioning isn't insurmountable. Just makes it a bit more tricky and more complex to keep track of everything.
18:47:29 <pnorman> I think there are some techniques that might help.
18:47:45 <pnorman> On a different matter, anything we want to bring up with the new board?
18:58:02 <zere> i don't have anything.
18:58:40 <zere> also, probably best for the new board to get at least one meeting under its belt before putting more stuff on its agenda.
19:06:18 <zere> RichardF: ping?
19:06:25 <RichardF> zere: pong
19:07:21 <zere> i'm playing around with the routing branch, but getting "cancelled" XMLHTTPRequests to router.project-osrm.org because there's no access-controll-allow-origin headers. is that something you've seen?
19:08:09 <RichardF> zere: curious - no, I haven't seen that.
19:08:14 <RichardF> let me look...
19:08:57 <apmon> zere: Are we going to get the routing branch merged?
19:09:45 <RichardF> zere: request seems to go through ok for me
19:10:02 <RichardF> e.g. http://jsrouting.apis.dev.openstreetmap.org/directions?engine=osrm_car&route=51.5830%2C-2.1022%3B51.7522%2C-1.2582#map=9/51.6427/-1.6795
19:11:58 <zere> ah
19:12:01 <zere> that one works
19:12:11 <zere> perhaps it doesn't add ACAO on failures?
19:13:27 <zere> RichardF: okay, that's not it. i have no idea. thank you, whatever you did, it now works :-)
19:13:33 <RichardF> heheh
19:13:38 <RichardF> I did nothing but... good :)
19:13:49 * zere assumes charlbury is now down one black rat
19:14:03 <zere> because sacrificing a black rat is the same as a black chicken, right?
19:14:16 <RichardF> believe it or not, the Rose & Crown doesn't sell Black Rat any more!
19:14:34 <RichardF> instead they have a rotating guest cider which is generally nicer. also which frequently causes me to rotate.
19:15:20 <zere> dustfield farm? interesting reverse-geocode for the rose and crown
19:16:15 <zere> anyway, getting back to the point...
19:16:35 <zere> apmon: yes, i'm trying to fix up a bunch of stuff and TomH will be reviewing shortly, i believe
19:17:12 <zere> to be honest, i'm not sure what i'm doing right now is making it significantly better-looking... but *shrug*
19:18:12 <RichardF> I have managed to b0rk my local git somehow, again, but I'll take a look when I can :)
19:18:34 <apmon> Oh, that would be awesome!
19:20:11 <apmon> Is there anything I can help with on that?
19:20:32 <zere> RichardF: i hope to have a PR for you later on
19:20:45 <RichardF> zere: excellent!
19:24:54 <zere> apmon: nothing right now, i think... but i'm sure there'll be some stuff soon, when the review gets going
19:25:40 <zere> btw, i notice that TomH pushed the overpass query stuff that he had been working on live - did that get announced anywhere?
19:27:06 <pnorman> no
19:27:37 <pnorman> Well, I announced it last week at a local meetup, but that doesn't count ;)
19:34:03 <apmon> Great to see these nice new features.
19:34:37 <zere> i wonder if it was intended as a low-key soft launch...
19:35:11 <zere> i guess it ended up that way anyway. perhaps i should check with TomH, and ask someone to write a blog post if it's not a soft-launch feature.
19:35:38 <zere> #action zere - find out if the overpass query is supposed to be a soft launch. blog post if appropriate.
19:35:55 <zere> anything else, or are we all done for this week?
19:36:39 <apmon> Depending on how soon the routing feature would possibly be ready for merge and if things are supposed to be a soft launch, those might be good candidates to try and get some wider press coverage?
19:38:47 <apmon> One would probably have to be careful of how to formulate things to not appear that OSM is lagging 10 years or so behind google maps, but it could hopefully garner quite a bit of press coverage.
19:42:00 <zere> yeah, if we're ready to publicise them, i'll poke CWG and see if they can get some press interest.
19:42:16 <zere> thanks to everyone for coming. hope to see you next week!
19:42:24 <zere> as pnorman pointed out, at 17:30 UTC :-)