We didn't do a mapping evening in a while so before I go any further...

Mapping Evening Tonight and Mulberry Bush pub from 8p.m.

The last mapping evening was much displaced in temporal and also geospatial dimensions. At least it felt like quite a long bike ride for me. But that's probably because I got a bit lost on the way to Pimlico. I still had time to zip around and pick up some building details before heading to the pub.

This pub meet-up was image of the week!

Quite a nice happy photo, although we actually had more people turning up a bit later, and more sunshine a little earlier. I think we can try for a better photo (maybe tonight??) Anyway... sitting outside the pub there, we talked about....

An amusing google maps screw-up. They had bizarrely moved Vauxhall station to the wrong side of the river! Quite a serious navigationally confusing mistake for a them to be making to the transport network of a major world city. They've since fixed it, as you might expect, when a few people like me start taking the mick out of them (and recommending people "use a better map") but in fact this has been the first in a string of tube station screw ups. Camden Town tube station has gone missing (pretty major intersection of the Northern Line) and Southwark Station is written in Japanese. They'll probably fix these things some time in the next few months, but they probably won't fix other details, like the failure to show up some parks in green. So the message remains the same "Use a better map" Use a map where details can be added and problems fixed quickly by local people who care.

While talking about this, we did also speculate about how and why google would get Vauxhall station on the wrong side of the river. Clearly it's some kind of algorithmic auto-positioning based on information gathered from ... somewhere. I was imagining it was done based on web crawling (like their search results) See old blog post from Ricard for more g-speculation prompted by the old Argleton mystery. Or maybe they look at people placing markers on embedded google maps around the web (which in fact they would not need to "crawl", they might be able to just mine their map API access data) Matt had another suggestion. Maybe they gather data on users adjusting routes. So if many users search for a route to vauxhall and then drag the marker to the other side of the river, that might start to count as a correction. Clearly all of these approaches would involve some checks and balances averaging over several repeated inputs. We know they do web crawling stuff for more minor POI data. Guess they're putting a bit too much faith in it, when they do it with tube stations.

And somebody suggested some kind of mischievous google bomb experiment to test the theories. Maybe I'll come back on that idea :-)

Alex talked with Matt about Matt's Terracer plugin and some problems and quirks of its geometry calculations (Alex since did some work on it at the hack weekend)

Ollie was talking about CASA work, such as the boris bikes animation video. He does a lot of stuff with the 'Processing' library, which is pretty neat for graphics, but not geo-aware at all, so you have to convert things to x-y coordinates on input.

Тюмень (Tyumen) is a relationtastically mapped city in Russia. Relations with is_in tags. So that's doubly messily pointless.

We had Azmat along, who is the creator of the rather neat busmapper.co.uk. I've been doing some work with him on that. The journey times you see appearing after you've found some routes. They're being loaded from placr.co.uk bus timetable data (via an API) Azmat was interested in meeting more geo-folk, but I guess we need to work on persuading him to switch to OpenStreetMap.

Andy told us about secret military installations near where he's from in Scotland. I cheerily pointed out to him that if world war 3 breaks out, he'll either get nuked while living in London or while back home near our missile stockpile. But we got onto that because it's interesting how some of these things are missing from OS maps, and sometimes even hidden by fake clouds in aerial imagery.

And speaking of aerial imagery, we talked about Bing offsets again.

Finally we talked about Derick's blog post about using OpenStreetMap, and about the MapQuest static maps API. Static maps (dynamically generated static map images) are really quite neat as a way of making maps easy to embed on websites. A long time ago I made a MediaWiki extension to embed static maps (linked to the osm.org homepage). I prefer this to the other one I made, which embeds a slippy map with OpenLayers. Slippiness is overrated. One problem with static map images though, is that it feels like you're doing something unstable, pointing an IMG tag at an external service which is doing the work of stitching tiles on the fly. That's not helped by dodgy looking URLs involving the word 'dev', so it's good that MapQuest are providing a more stable looking service. But we got talking about CMS integration e.g. plugins for wordpress, drupal, etc. Actually in any situation where PHP is available it's easy enough for tile stitching caching to be done "locally" (on more numerous servers where CMS/plugins are running) This has a couple of advantages over the centralised service approach. If anyone fancies hacking on something like that, it could be worth following the discussion of standardised request parameters.

We had some new people along, having a go at mapping for the first time. Hopefully we didn't scare them off with all this techy talk.

If you like the sound of this techy talk, or just some friendly beers (we do have some more straightforward conversations too) come along tonight! Muberry Bush pub from 8p.m..

Fancy some mapping beforehand? It's beautiful weather for it! I was planning to do a MapCraft based cake diagram this time. Should be available here but it seems to have gone offline, so I'll have to set it up as a boring old wiki-based sign up. Grab yourself a cake slice if you fancy it!