As I explained previously there's a small area of Islington around Pentonville Road and north of Clerkenwell, which is sort of sandwiched between Blumpsy's blob of mega building detail and the main central London building coverage, so we headed there for last week's London Mapping Party.

I was conscious that this should really be my territory. In fact I mapped a slice of the cake and also off-cake cake up Amwell Street, which is actually the route I cycle home every day. I was in two minds about this. Feeling guilty that I hadn't managed to map it fully before, perhaps I should've left it to map during my cycling commuting some other time, but... it would certainly annoy me if somebody else mapped it first! I have added one or two details week by week from my cycle to work, but I'd have to be more organised (and cycle considerably more slowly) to do it properly. So last week I was on my usual cycle route home, but this time taking photos of every shop and other such details.

This process really highlighted for me the crazy serendipitous psychological state of mind I enter into when I go mapping. Despite being a very familiar road, it's not until I scrutinise it for OpenStreetMap that I realise things like this:

• On that road there's a remarkable number of stylish looking shops with seemingly no name. How do you run a business like that???


• There's a shop (or perhaps an office) offering dog walking services.


• There's a gap in the buildings with big three metre high construction hoardings, but then bushes growing over the top. The land must've been unused/available for some time. Surprising for central London.


• And finally there was a massive ugly 60s building bristling with security fences and surveillance; a high security office or perhaps a prison or something, also with no clear name, but I noticed the facility (whatever it is) is operated by serco. Very mysterious.

I love discovering and recording this kind of stuff, and these are all things I never noticed while cycling past every day.

I'm disappointed to discover from wikipedia that The Angel Islington pub is not the same building as "The Angel" wetherspoons pub we were in. The original inn was on the corner. The current pub is nextdoor. Rubbish. ... Still it has bags of character as I'm sure you can imagine of a wetherspoons.

In the pub we talked about offline maps in mobile apps. Dan told us about a friend who was seeking offline mobile map technology for the police service. A popular option for this is offmaps. It seems to me that offmaps 2 is hobbled version offering only pre-packaged cities, while the original offmaps is more flexible. Weirdly it's still available but more expensive. Perhaps it's to do with problems coping with the tile traffic it generates. Offmaps hits CloudMade tile servers rather than the donation funded OpenStreetMap tile server (which is good) but it seems like the really smart way to go for offline maps ....is vector! Mapdroyd delivers on this brilliantly in my opinion. I'm not being paid to say this or anything. If you have an android, get mapdroyd. It's the single most useful app I have on my phone. Maybe I find it so useful just because I'm a cheapskate with topping up my data allowance. It's also perhaps less useful for people who don't enjoy reading maps as much as I do, since it's missing a search facility. I'm more annoyed about them omitting pub icons. So despite being awesome there's room for improvement.

None of this was directly answering Dan's friend's question because for the police service they wanted to code their own app functionality, but include offline maps (from OpenStreetMap we hope!). So they need offline maps as a drop-in programmable component. We weren't sure what the best recommendations were. Looking at it now I see the droyd SDK might be just the thing, for a price, but I'm sure there are free and open source implementations of tile map offline storing, and presumably various attempts at offline vector maps.

I whipped out my iPad to show off UK Travel Options. This led on from the previous conversation because it's also vector based. It's nicer looking (whizzy 3D) and shows more POI details than mapdroyd, but it's not designed to work offline. In fact it does work offline for a reasonable chunk of map (a few cached data tiles) but mostly works better with an internet connection. The whizzy 3D is made extra whizzy by the new support for multitouch rotating. If you have an iphone/ipod touch/ipad, try it. It's free! It's particularly gorgeous on an ipad.

I had a long chat with Andy Allan about various assorted big topics. We started on public transport mapping and stations represented as nodes vs areas. This lead on to general principles of "mapping what's on the ground" vs mapping for data users, and the need to do less of the latter, by having tools to do smart post-processing of OSM data for data users. This lead on to a "level of detail" discussion. Extending the "mapping what's on the ground" argument to extremes leads to excess detail mapping. Things like mapping sidewalks and mapping roads as areas. These things are to be avoided in my opinion. Ease of use of data is perhaps not a legitimate argument against this (Area roads could be post-processed to be be linear) but when we push towards silly levels of detail this sends confusing messages to new mappers and impacts our ability to achieve good and even coverage. Then again it's all relative. It could well be argued that I'm doing exactly this when I promote building/POI mapping in central London. ...Big topic

...but that wasn't all. We got onto the big topic of wiki documentation. Andy complained that there's really very few pages of the wiki which offer a good tidy overview of a topic suitable for linking to for beginners. It's a big sprawling mass of information. It can be painful to imagine a newcomer's journey into this morass. Most wiki pages are heavily interlinked, which is probably a good thing, but starting on the well written page you intended them to read, a newcomer can quickly stray into some out-of-date defunct badly written cruft. Even the well written pages get gradually less well written over time as people chip in with contributions which are well-meaning, can be factually accurate, but ultimately detrimental. Andy felt that forking documents git-style could offer a better approach. I think there could be something in that, or in the general idea of separating off some documents as more "fixed" or "curated". The wiki does particularly badly at things like the "beginners guide" where it would better be written by one knowledgeable person with one vision of what to focus on, and how to spin the narrative. It's almost like we need a competition. Who can produce the best three-page PDF? On the other hand, I also think that the problems of the wiki can be overegged. It's easy to blame the underlying software/procedures but ultimately there's a certain amount of "hard work" of documentation, which will always need to happen to produce good results regardless of the platform.

We talked about how we've got lots of new hardware. Nine new machines I think we said (can't read my own writing). This will either mean OpenStreetMap gets an amazing new array of applications running and solving all problems, including world hunger and curing cancer through the power of OpenStreetMap, or we'll get no new applications, but OSM will do what it does in a faster and more scalable and reliable manner. Either way it's good news.

With that we headed for burritos. I had no idea there was a burrito shop nearby, but there it was. node:348407534. Bonus!

I've just set up details for the next London mapping evening It's this coming Tuesday in South Kensington. Don't forget to invite people along if you know anyone in the area who might be interested. Spread the word!