The next London pub meet-up will be this Thursday at Ye Olde Cheshire Cheese.

Last time we had a pub meet-up at the blue posts, where we got a table in the very quiet upstairs area. All the better for map chat.

Martin from cyclestreets was there. He showed us a couple of exciting cyclestreets initiatives. They've funded the development of (hired Andy Allan to create) a feature of Potlatch 2 for merging in tags from a vector dataset. This should be generically useable for a range of "community import" type activities (having a parallel dataset as a source and selectively bringing in data to OpenStreetMap) but the work was specifically aimed at letting us use a cycling dataset from the DfT and to bring in extra tags giving things like the widths of cycle lanes. Clearly this is not data we can just automatically import (the approach some folks will all too quickly reach for). This P2 feature lets us be more careful and selective, blending new information with our locally gathered data.The data is shown as a seperate layer with geometries appearing as very wide lines which can be selected.

Martin also showed us a new brochure they had been working on which looked pretty cool. It had a cycling slant in some of the introductary pages, and a Scotland slant too, but mostly (as intended) it will work well as generic OpenStreetMap promotion leading into beginner documentation on how to contribute to OpenStreetMap, with some nice real-world photographic examples.

We talked about some upcoming syadmin challenges. The current database server 'smaug' has developed some faults. This coupled with some other considerations, will mean we need to move the main database onto the new server. Moving the data across is a bit tricky. The only safe way to do it will be to put things in read only mode for a period of a few days while the data is copied across as a pgdump piped across to the other server. I wondered whether there might be any way of avoiding the read-only period, but there isn't really. The problem is the database is not just the geodata. It's users and sessions and things like this. We'd also want to be moving to a later postgres version. It won't be possible to ask the two machines to do "clustering" because of the version mismatch. There will soon be some announcements about downtime for this work (By the way, this is me "backseat sysadmining". I actually have no idea about this stuff and I'm glad it's not me doing it. For details you'd have to ask the sysadmins)

They were having some downtime in the kitchen of this Blue Posts that night, so we had to make alternative food plans. Even though Matt wasn't there we went for Burritos, so I made sure I got some juicey burrito photos to taunt Matt with:

Over burritos we got the rather excellent news that Foursquare did a switch2osm!. Foursquare are big. 15 million users. And although they only switched their website, not their apps (which many users experience foursquare through most of the time), that's still a great boost to the number of users seeing our maps. That's what we do it for! They're using a MapBox tile server, so presumably this will pick-up changes and re-render. The custom map style does of course not show any OpenStreetMap POI data, that's where things would get a bit confusing for all those foursquare users who are busy building and maintaining Foursquare's closed database of "check-in venues". We can hope though, that some small percentage of foursquare users will look into contributing to OpenStreetMap as a result of this switch. Great news!

All of this spurred another discussion about Google Map Maker and features for fixing google's POI locations (in the UK there are features to do this, but not going by the "Map Maker" name). Some of those eating burritos had experimented with google map making (traitors!) and wondered how the verification works (and why bother) when some contributions are based on local knowledge, with nothing in the aerial imagery to back them up. Also they noticed variations in the the length of time it took to have changes verified from a couple of months to just a couple of days, perhaps depending on some trust metrics, or perhaps whenever the indian map maker minions get around to it. Google have a lot of paid indians doing map maker moderation. They also have a lot of unpaid indians contributing to their closed google-owned data. Google should use OpenStreetMap. All the unpaid Indians definately should. switch2osm!

Here's a future pub discussion we should have: Alex has put forward a plan for an armchair mapping initiative related to building outlines in central London. OSMLondoners please have a read and comment. Obviously this leads into the broader discussion of "armchair mapping" in general. I recently tried to lay out a wiki page on 'armchair mapping'. At the moment I've heard strong criticism from pro-armchair mappers who say this is too negative. But I beleive it to be a sensible set of guidelines and an even reflection of some opposing viewpoints on the matter. Tell me what you think on the 'Talk' page there. But how does all of this apply to the London building outlines idea? A good one to chat about over some beers perhaps.

And speaking of building outlines some people are getting itchy mapping feet as the daylight is coming back in the evenings now. I didn't pick a pub near a massive amount of mapping, but there is a sprinkling of data which needs re-mapping in this area (bad map OSMI) Quite a few building outlines among other things. This is fortuitously located around the neighbourhood of Ye Olde Cheshire Cheese. Join us back in those cosy caverns on Thursday! All the details on the wiki page.