Working Group Minutes/EWG 2013-10-07


IRC nick Real name
gravitystorm Andy Allan
iandees Ian Dees
RichardF Richard Fairhurst
TomH Tom Hughes
zere Matt Amos


  • Review of 2013 goals
    • Didn't spend any of the hack event money. Was it because they didn't need it, want it or know about it?
    • ACTION gravitystorm to ask event organisers if they knew about it.
    • ACTION zere to approach CWG about publicising it, even just for the small remainder of the year.
  • Developer documentation for many projects isn't great. (i.e: Docs about how to understand a project and start modifying it, rather than just how to run it.)
    • Could run a "doc-weekend" perhaps.
    • Issues with finding information on the wiki - would be better all in one place.
    • ACTION zere to run a short survey to find out where documentation and development is thought to be most needed.


17:01:14 <zere> minutes of the last meeting: please let me know if anything needs correcting
17:01:33 <zere> #topic actions of previous meetings
17:02:01 <zere> pnorman: how goes it with the switch2osm carto / AWS postgis dumps stuff?
17:05:33 <gravitystorm> maybe pnorman is busy elsewhere
17:05:47 <zere> it seems so. will will come back to it.
17:05:53 <zere> #topic 2014 planning
17:06:07 <zere> i wanted to talk about planning for 2014 a little bit
17:06:12 <TomH> I think we should plan on 2014 happening, yes
17:06:24 <zere> 2014 is so hard to type - my fingers want to type 1024 instead.
17:07:48 <gravitystorm> zere: what sort of things to you have in mind for EWG in 2014?
17:07:49 <zere> looking at the 2013 plan, we had an allocation of $500 over 10 events to subsidise for development. a total of $0 of this has been claimed.
17:08:29 <zere> which implies that it's either unnecessary, or we're not doing a good job of letting people know it's available.
17:08:57 <gravitystorm> Unnecessary would imply that we have outcomes in mind, and those outcomes have all been met
17:09:09 <gravitystorm> so first off, we'd need to know if 10 events happened by themselves
17:09:58 <gravitystorm> I think your second hypothesis is probably true, regardless of the above
17:10:39 <zere> i count 10
17:11:25 <zere> there was one in berlin, chicago, glasgow, london, toronto and vienna. two in karlsruhe and one after each of SOTM and SOTM-US.
17:12:05 <zere> however, it's debatable whether those 10 events might have been 20 if $500 would have been of any use to the organisers, and they were aware of it.
17:12:34 <gravitystorm> was the budget $50 or $500 per event?
17:12:55 <zere> $500 per event, $5,000 in total for all 10 events.
17:13:16 <zere> well, $5,000 in total, of which we expected to be able to subsidise 10 events.
17:13:46 <zere> i guess if people had wanted less than $500 it would have been possible to subsidise more.
17:13:51 <gravitystorm> sure
17:14:20 <zere> one item was "Make information about how to apply for the subsidy more widely available via MT &
17:14:23 <zere> CWG.
17:14:25 <gravitystorm> I can volunteer here for some research then - I'll ask each of the organisers if they knew there was money available
17:15:04 <zere> #action gravitystorm to ask hack event organisers if they knew there was a subsidy available.
17:15:07 <zere> gravitystorm: thanks :-)
17:15:37 <zere> i guess the next question is: should we at least try and subsidise some events before the end of the year?
17:16:22 <zere> i'm happy to reach out to CWG and see if we can get a blog post up about it. only 83% of the year gone, but might still be worth a try.
17:16:57 <gravitystorm> sounds good to me - so long as the money is going to good use, we should spend it
17:18:00 <zere> #action zere to reach out to CWG to publicise the availability of the subsidy.
17:18:23 <zere> ok. other goals we had:
17:18:32 <zere> "Improve documentation for rails_port (READMEs, installation docs, developer docs) with
17:18:41 <zere> the aim of reducing the barrier to entry for new developers and 3rd­party users
17:18:44 <zere> "
17:19:34 <gravitystorm> I think we did that, in as much as they got a bit of a rewrite but are at least now in the right place
17:19:59 <gravitystorm> there's a couple of dangling issues with that, chiefly the population of a database using an extract, but they at least are in github
17:20:44 <gravitystorm> Are there other projects that need some documentation TLC? Getting started docs are pretty much as in-scope for EWG as things get.
17:21:23 <zere> "Improve documentation, and access to documentation, for other software used in the wider OSM system."
17:21:26 <iandees> as organizer of the chicago event, i can say i don't think i knew the subsidy was available
17:21:52 <iandees> i spent a few hundred bucks on food and it was covered by event sponsors (who also paid $2000 for the room for two weekends)
17:22:21 <zere> iandees: hmm... do you think if you'd have known that it would have helped? or was it not an issue when you'd found sponsors?
17:22:57 <iandees> i think for those that have a location but want to run a long event (thus requiring food support) the $500 would help a lot
17:23:12 <zere> as for the "improve documentation ... wider OSM system" - i guess the switch2osm stuff falls partly under that, but otherwise i'm not aware of any other doc-writing we've been involvd in.
17:23:39 <zere> iandees: thanks. still worth pushing it, then. good.
17:24:02 <iandees> we (some osm us folks) did apply for the Google Doc Summit thing but did not get accepted
17:24:35 <zere> perhaps we need a doc-weekend ;-)
17:24:45 <iandees> it was surprisingly effective
17:25:21 <iandees> most of the people that were at the one in 2011 on google's campus were very sceptical when it started but ended up with ~200 page books that were pretty darn good at the end
17:25:29 <iandees> not that we have to write another book
17:25:45 <iandees> but a group can produce/fix a lot of docs in a weekend
17:25:51 <zere> was there enough "parallelism" to keep multiple people occupied, or is it more of a collaborative "pair-documenting" type thing?
17:26:07 <iandees> we ended up doing chapters at a time in parallel
17:27:04 <zere> do we think "Improve routes to getting help and advice when installing, setting up and using OSM­related software." has been tackled?
17:27:59 <iandees> i still get the feeling that the wiki is a minefield for newcomers
17:28:07 <iandees> but stuff like switch2osm is an excellent resource
17:28:25 <zere> not just for newcomers - i find it tough to navigate too
17:29:32 <iandees> it would be extremely hard to do, but i think a repo that accepts pull requests and bugs for docs (rather than anyone-can-edit wiki) would help for docs
17:29:37 <zere> would it be worth collecting stuff on a development-specific site similar to or
17:29:40 <iandees> extremely hard because our software is so diverse
17:30:38 <zere> indeed, and lots of it is changing rapidly at one end of the spectrum, or is basically unmaintained at the other.
17:30:57 <gravitystorm> I think there's room for a bit of "where can you find all the bits of software" and "how they all fit together", but I'm very keen on making sure project-specific docs live in the project's own repo
17:31:36 <zere> sure. that's fine - but the presentation of them doesn't have to be scattered around
17:31:42 <gravitystorm> so I could see a "development overview" site replacing all the development-related information on the wiki, but not much more than 5-10 pages of material
17:32:05 <iandees> is there a pattern to what kind of technical requests we get outside of "i want a tile server"?
17:32:37 <zere> "why am i blocked from service X"?
17:33:09 <RichardF> iandees: we're starting to quite a lot of "how do I set up Nominatim", and to a much smaller extent "how do I set up OSRM"
17:33:17 <RichardF> s/to quite/to see quite/
17:33:51 <iandees> maybe it would be good to start sections in switch2osm for those two?
17:33:53 <zere> but "how do i set up X (in order to use it)" is quite different from "how do i set up X (in order to modify it or contribute to it)"
17:33:59 <gravitystorm> zere: indeed
17:34:28 <zere> i think switch2osm would be perfect for the former. and i dunno about the latter
17:34:46 <RichardF> iandees: yes, I think so. I could perhaps do OSRM but I wouldn't have the first clue about Nominatim
17:34:47 <zere> certainly the current resources (wiki, github, etc...) can be hard to navigate.
17:35:23 <RichardF> zere: yes, absolutely. moving "how do I help development" onto s2o would be daft
17:35:44 <zere> RichardF: i'm sure we can convince woodpeck to help with nominatim, if no-one else volunteers. or i can twist twain47's arm a bit.
17:35:52 <RichardF> that'd be great.
17:36:52 <zere> one possibly interesting, or possibly tangential, idea is vagrant. anyone used that? found it useful?
17:37:06 <iandees> use it with osmdevbox all the time
17:37:08 <iandees> works great
17:37:59 <gravitystorm> I use vagrant all the time.
17:38:16 <zere> if we had a concerted effort to get a Vagrantfile for OSM projects, do you think it would help people get started?
17:38:18 <iandees> is the repo. check it out, vagrant up, vagrant ssh and you've got yourself an openstreetmap-website going
17:39:37 <iandees> not sure if one vagrantfile for them all would work very well, but getting one for each might be useful for people to experiment with
17:40:16 <zere> yeah, sorry, i meant that as well as a README(.md) in the top of each repo, we should have a Vagrantfile too
17:40:54 <zere> one Vagrantfile for everything might even be impossible, depending on the various versions of the software dependencies that things need...
17:41:01 <gravitystorm> iandees: I hadn't seen that before. We could probably do similar things using the actual chef recipes
17:41:02 <iandees> yea
17:41:40 <zere> but the question is: would it help? do you do development within Vagrant, or is it mainly for testing or what?
17:41:48 <zere> i admit to complete ignorance about it.
17:42:04 <zere> e.g: is it watching your source files for changes?
17:42:10 <gravitystorm> So I don't use it for development, only for testing chef recipes.
17:42:23 <iandees> i tend to use vagrant (and virtualbox in general) when i don't want to attempt setting something up because a) it's hard/impossible on a Mac and b) i don't want to screw up my environment with dev-ey things
17:42:48 <gravitystorm>
17:44:10 <gravitystorm> yeah, it can be used as a command-line-access-to-virtualbox kind of thing, if nothing else. I worry that if we pointed people towards vagrant for e.g. developing the rails port, that's just another level of complexity
17:44:30 <zere> iandees: how does the dev stuff work? do you ssh into the VM and edit stuff there, or do you edit outside and keep doing "up"s?
17:44:50 <iandees> i ssh and edit stuff in there, but it's possible to share directories between the VM and host
17:45:08 <TomH> well once it's up it is just a VM like any other
17:45:15 <TomH> vagrant is just a provisioning tool
17:45:39 <TomH> AIU
17:45:52 <iandees> yep. it's really just a better interface to virtualbox than the VirtualBoxAdmin command line
17:46:14 <zere> sure. i'm just wondering how it gets used in practice, whether it helps to do anything more than "rake test" in vagrant, etc...
17:47:16 <zere> at the end of the day: whether it's worth making this a goal for ourselves in 2014, with the aim being that it helps people to get started more easily.
17:47:17 <iandees> because 'up' does the whole OS boot process, it's pretty time consuming
17:48:00 <iandees> i think it might be more effective to make consistent, good documentation for setup on some commonly-used platform rather than use Vagrantfile
17:48:09 <iandees> for the projects that most need it
17:48:35 <zere> fair enough.
17:48:54 <zere> but that's harder than just writing a few Vagrantfiles ;-)
17:49:26 <gravitystorm> iandees: yes, that's my feeling too. For switch2osm things, I think VMs or Vagrantfiles (or lxc's or whatever else) might be useful, but for development I think clear instructions are better. Many devs will have their own prefs on dev environments anyway
17:52:25 <gravitystorm> the main question (for me) is which of the projects have the greatest (poor docs) x (need for developers) ?
17:53:22 <zere> osm2pgsql?
17:54:29 <zere> in terms of developer docs, that is. the switch2osm docs, and the ones pnorman is working on, are better than for other projects.
17:56:18 <gravitystorm> I suspect OWL too
17:56:18 <zere> actually, perhaps it's worth going through a bunch of projects to try and put numbers on "poorness of docs" and "need for developers".
17:56:29 <zere> list me some projects!
17:57:13 * gravitystorm looks at the non-yet-existant openstreetmap-developer-overview website we mentioned earlier
17:57:18 <iandees> :)
17:57:32 <iandees> didn't woodpeck have a neat openlayers-based presentation with a fairly good overview?
17:57:54 <iandees> we should adapt that and add popup boxes with dev and setup docs ;)
17:58:28 <zere> OMG - google docs now has eye-scratchingly awful themes... it really is the successor to MS Office.
18:08:59 <zere> gravitystorm, iandees:
18:09:03 <zere> any more to add to that?
18:09:48 <iandees> "poorness of docs" is probably a bad measure to use :)
18:10:05 <iandees> do you mean a "good amount of poorness"?
18:10:11 <iandees> :)
18:10:22 <iandees> maybe overpass
18:11:13 <gravitystorm> descriptions need more emphasis on it being *developer* documentation
18:11:18 <gravitystorm> maybe osmosis
18:13:46 <zere>
18:13:49 <zere> well
18:13:50 <zere> updated
18:13:53 <zere> how does it look now?
18:14:00 <gravitystorm> are we sticking to core(ish) software, or also mkmgap?
18:14:30 <zere> i dunno. happy to include it, but i don't want the list to get too long.
18:14:48 <gravitystorm> leave it for now then, otherwise we could add another 40
18:14:49 <iandees> who is the audience for this survey? it might be good to include a "what the hell is this?" option?
18:15:09 <gravitystorm> iandees: I'd say dev@ would be a good start
18:15:17 <gravitystorm> and #osm-dev, of course
18:15:49 <iandees> are we asking dev@ to go to the project and evaluate the doc or do it from what they know?
18:16:13 <zere> yes ;-)
18:16:17 <iandees> :) ok
18:21:28 <zere> hmm... i'll continue this after the meeting.
18:21:34 <zere> thanks to everyone for coming!