Working Group Minutes/EWG 2011-11-07

From OpenStreetMap Foundation


IRC nick Real name
apmon Kai Krueger
CGI173 Steve Coast
RichardF_ Richard Fairhurst
TomH Tom Hughes
zere Matt Amos


  • RichardF_ and Grant have made an intermediary page for blocked applications to go to, alleviating the pressure and discomfort in #osm.
  • In the first meeting of each month, EWG will help CWG to make a list of salient changes in the code. This combats the myth that "nothing ever changes".
  • Some discussion over whether providing VMs or an install script would be an easier route to getting everything rails_port needs installed correctly.
    • apmon to work on a script/package for getting dependencies installed.
    • There was some discussion of how to reduce the number of dependencies, but no clear consensus.


18:02 < TomH> evening
18:02 < apmon> Hello
18:02 < zere> last meeting minutes as always, let me know if there's anything objectionable in there
18:02  * apmon goes and reads them
18:03 < zere> it was pretty short last time, which hopefully limits the scope for mistakes ;-)
18:05 < RichardF_> hokay. since then Grant and I have JFDIed the "blocked tile client" issue :)
18:06 < zere> great! does it send people to a static / wiki page or something?
18:06 < apmon> yes
18:06 < apmon>
18:07 < RichardF_> and makes you feel guilty for exploiting volunteers, too ;)
18:08 < apmon> Anything on the agenda for today?
18:10 < zere> i'm a bit zonked out from all the international jet-setting, so i have to apologise for not setting an agenda. does anyone want to raise something for discussion?
18:10 < RichardF_> one thing that came out of SWG (and is best punted towards both CWG and EWG) is better communicating the changes that happen to the website, to dispel the impression that "nothing ever changes".
18:10 < TomH> zere: anything relevant from the F2F we should consider?
18:10 < RichardF_> CWG is happy to help. so I guess EWG ought to know that CWG would like to hear about cool stuff that's under development so that we can shout about it :)
18:10 < apmon> RichardF_: I was about to bring that up too... ;-)
18:12 < apmon> TomH: Is there anything ewg can or should do with regard to the rails3 transition?
18:12 < zere> perhaps each week we could go through the commit log and/or major branches and provide CWG a summary of what's happened and why it's important / useful?
18:13 < TomH> apmon: well I have identified the source of the extra memcache traffic, and disabled it pending validation of whether it is ok, but I don't think it caused the session issue
18:13 < RichardF_> zere: that'd be terrific.
18:13 < TomH> I have also added code to store the username in a cookie and cross validate the session, which is a bit of a band aid but shoud help
18:13 < apmon> Is it worth posting on one of the rails lists to see if anyone else has experienced anything similar
18:14 < TomH> the remaining concern otherwise is the extra cpu load
18:14 < apmon> or if they might have an idea what caused it?
18:14 < TomH> apmon: I doubt it really
18:14 < apmon> zere: I am not sure quite that much happens to warrant weekly updates, but otherwise the idea makes sense 
18:17 < zere> yeah, and maybe weekly updates is more than anyone wants to read. is monthly more appropriate?
18:17 < zere> we could do the summary on the first EWG of each month or something.
18:17 < apmon> TomH: What happens if the cookies don't match up? They just get logged out and an alert gets sent?
18:20 < zere> for the last month i can see: lots of PL2 updates, a UI fix to be able to click users in the "nearby" map, and some XSS/security type fixes. 
18:21 < zere> RichardF_: what are the salient points of what went into PL2 since beginning of last month?
18:21 < RichardF_> let me look...
18:22 < RichardF_> blimey. quite a lot. might be better if I talk to CWG directly.
18:22  * RichardF_ talks to self
18:22 < TomH> apmon: it just clears the session and acts like they weren't logged in
18:22 < zere> as for the F2F, yes, there's plenty relevant. but i think there's going to be a grand announcement, so i'm not sure that i should jump the gun here.
18:23 < apmon> F2F? Face to face of the osmf board?
18:23 < zere> yup
18:23 < apmon> when was that?
18:23 < RichardF_> this weekend just gone
18:24 < zere> i got the red-eye in this morning. or last night. or something. which explains why my eyes are red and my brain is dribbling out through my ears.
18:25 < apmon> Well, it will be interesting to read about it then
18:25 < zere> yes. yes, it will ;-)
18:26 < TomH> so if last time is anything to go by the grand announcement should happen in a few months time ;-)
18:28 < zere> that might be how long it takes to get over the jet-lag ;-)
18:28 < apmon> where was it?
18:31 < RichardF_> Seattle
18:33 < RichardF_> going back to a few weeks ago - where did we get on the issue of "making the OSM codebase easier to play with"? we were talking about VM images etc. - did we get anywhere?
18:34 < zere> not afaik. at least, not to the point of publishing a VM image.
18:34 < RichardF_> hm. but I think we decided that would be a nice-to-have, right?
18:34 < zere> the feedback that apmon got from the german mailing list seemed to indicate that VMs weren't as desirable as good install instructions.
18:35 < zere> but, of course, there's no reason we can't have both. :-)
18:35 < apmon> Possibly something like the mod_tile/renderd ubuntu packages would be best.
18:36 < RichardF_> that would be really good, yes.
18:36 < apmon> For those who have a linux installation already, they could just use that and others would install their own VM and install the packages there
18:36 < RichardF_> yep. that works.
18:37 < zere> so that's basically debs for the gems that we use, right?
18:37 -!- CGI173 [] has joined #osm-ewg
18:37 < apmon> It could be a deb that calls a "postinstall" script that installs the gems the normal way
18:38 < RichardF_> I guess so. what tends to throw me is that you find x requires y, or j conflicts with k, or l can only be installed in version n, or whatever.
18:38 < apmon> It would probably mostly be copy and past the decent install instructions from the wiki into a script and package that together with the dependencies as a deb
18:38 < RichardF_> having a script to do all of that so that you can Just Start Hacking would remove a whole bunch of heartache.
18:39 < apmon> I probably could give that a shot and see how far I get
18:39 < RichardF_> that'd be really terrific. I'd offer help but by definition I wouldn't be much use ;)
18:40 < apmon> Those packages (just like my mod_tile packages) won't meet the debian standards for inclusion, but they can be hosted on a ubuntu launchpad ppa
18:41 < zere> i haven't done this from scratch in a while, but what's needed other than rake gems:install?
18:41 < apmon> Although I am also trying to cut those packages down to try and push for inclusion into the proper repositories of debian and ubuntu.
18:42 < RichardF_> zere: install postgres, set up user and databases, install btree_gist stuff, set up first user, etc.?
18:42 < apmon> zere: So far the rails port instructions on the wiki have always worked pretty well for me.
18:44 < apmon> It is just a bit annoying to having to copy and past all the different bits from the wiki into the shell.
18:44 < RichardF_> what would be ideal is if someone technically capable and willing sees something they'd like to fix on; installs the package; patches the code; pushes patch back to us. should really be that easy.
18:44 < apmon> They will probably also need to be adjusted for rails 3 now, as a couple of things have changed
18:45  * RichardF_ nods
18:46 < apmon> As I have the rails 3 port running, I will just do it straight for that
18:46 < RichardF_> that'd be hugely cool :)
18:47 < apmon> and hope that we are close enough to redeploying that by the time I have done it, rails 3 will be running on the main site...
18:47 < zere> i guess we've gone too far towards postgres to make a sqlite backend feasible
18:54 < zere> i think we've discussed this before, but my brain is on standby right now: are there fallback pure-ruby libraries we could use for xml and image stuff where it's tough to install libxml & rmagick?
18:54 < TomH> well ruby has rexml builtin
18:54 < TomH> but it will have a different interface so is hard to substitue
18:55 < TomH> there are lots of other native gems we use anyway
18:58 < zere> i can only think of libxml, rmagick and pg off the top of my head... i thought the oauth, i18n, httpclient and so forth were all pure ruby?
18:59 < TomH> nokogiri
19:00 < zere> ah, it's not listed in environment.rb...
19:00 < TomH> it's a dependency of sanitze
19:04 < zere> looks like nokogiri might well be a superset of libxml functionality. maybe at some indeterminate point in the future i'll see what, if any, downside there'd be to removing the libxml dependency.
19:05 < zere> RichardF_: do you have what you need for CWG? want some explanation to go with the code changes in rails_port?
19:06 < RichardF_> zere: if you've got a chance to drop me and Richard W a line with three sentences on the code changes that'd be cool. CWG is in (looks) two hours :)
19:07 < zere> nop roblem% yuo canf ix my dodgt typong and grammer, yah?
19:08 < RichardF_> I can have a go ;)
19:08 < zere> awesome. is there anything else to discuss today?
19:10 < RichardF_> not from me
19:11 -!- apmon [] has left #osm-ewg []
19:13 < zere> cool. thanks to all! same time next week :-)