Working Group Minutes/EWG 2013-04-15

Attendees

IRC nick Real name
apmon Kai Krueger
pnorman Paul Norman
TomH Tom Hughes
zere Matt Amos

Summary

  • Carto style port:
    • apmon checked in a commit so that render_speedtest spits out per zoom layer rendering time information, e.g: [1]
    • ACTION pnorman to try benchmarking on benzene with existing tools. - pending
  • Notes branch:
    • No blocking issues remaining. Expected merge soon.

IRC Log

17:02:20 <zere> minutes of the last meeting: http://www.osmfoundation.org/wiki/Working_Group_Minutes/EWG_2013-04-08
17:02:30 <zere> please let me know if anything needs updating in there.
17:02:48 <zere> there's a couple of actions in there, so let's start with any news on those.
17:03:04 <zere> apmon: any news on render_speedtest improvements?
17:05:27 <zere> hmm... maybe he's afk. no matter.
17:05:39 <zere> pnorman: any news on benchmarking with existing tools?
17:07:19 <pnorman> I have the db imported locally, but i've been busy with swapping drives and other stuff
17:08:00 <zere> no worries.
17:08:12 <pnorman> anyways, afk, got virtually no sleep last night
17:08:21 <zere> was there anything more to discuss about the notes branch, TomH?
17:09:16 <apmon> zere: Well, render_list can kind of do it
17:09:33 <apmon> I checked in a commit yesterday, so that it spits out per zoom layer rendering time information
17:11:50 <zere> but it would still be a particular bbox it was rendering, so likely to nor stress the db as much as a "realistic" test?
17:13:42 <apmon> It will give the following output at the end: http://pastebin.com/yRRjvDsX
17:13:56 <apmon> zere: You can give a list of tiles to render to render_list
17:14:16 <apmon> so you can use a standard list for benchmarking e.g. derived from the real output of yevaud
17:14:50 <apmon> pnorman converted your log output from yevaud to a render_list compatible tile list
17:14:55 <zere> nice. what machine is that running on?! 19200 tiles in 22s!
17:15:09 <apmon> A machine with an empty DB... ;-)
17:15:35 <apmon> the 19200 are real tiles so 64 times more than metatiles
17:15:40 <zere> yeah, i wish i could get even close to the 13 metatiles/s that's showing on a single machine ;-)
17:16:55 <apmon> I haven't tried running it against a full db yet.
17:16:57 <TomH> zere: don't think so
17:17:28 <apmon> TomH: Is it ready to deploy then?
17:17:32 <zere> cool, are we close to a merge? anything that anyone could help out with?
17:19:06 <TomH> I'll probably do a final call on it this week
17:19:37 <apmon> Nice!
17:19:49 <zere> yep, very nice :-)
17:23:51 <apmon> zere: Also, I have tagged a new release of osm2pgsql last night.
17:24:01 <apmon> So hopefully I will get around soon to merge your patch
17:25:14 <zere> cool. and hopefully now the homebrew guys will get osm2gsql in their distro.
17:28:09 <zere> did anyone else have something they wanted to discuss?
17:30:18 <apmon> Was there still stuff regarding plans and budgets that needed discussing?
17:30:55 <zere> budget / plan was submitted, we should hear back after the next MT meeting - the 23rd, iirc.
17:38:24 <zere> ok, i guess it's a quick meeting today. thanks to everyone for coming!