Blue Posts meet-up last week

Posted by Harry Wood on 16 April 2011 in English (English)

Over on this talk page I've explained the predicament I'm in. It's no different from the end of last summer, but now we're at the beginning of a new mapping season. Where to go mapping? This and shortage of time meant I failed to prepare any cake diagram for Wednesday's london summer mapping party

As usual we kind of failed to make any solid plans in the pub, we did have some discussion of the topic. Derick was suggesting we should gather missing tube entrances (railway=subway_entrance nodes). I suggested these tasks (stations where there are none) could be added as openstreetbugs. There's lots of mapping to do based on fixing of London OpenStreetBugs, geofabrik fixmes, and keepright bugs, but somehow these tools don't seem to have a good enough throughput of useful bugs to make them very worthwhile. I think particularly in the case of the human reported OpenStreetBugs, if we could get input from more non-mapping new users, we might start to see more of an interesting turn-around of new bugs getting reported then fixed. Not entirely sure what they're missing, but they're missing something. They're like a communication channel which hasn't gained critical mass yet.

We also talked about 'source' tags. I've been wondering what these would look like mapped. Actually ITO's OSM Mapper lets us do this in a basic way:

But this is complicated these days by the fact that many people (myself included) have stopped putting source tags on ways and nodes, and instead put a source tag on changesets. That doesn't show up on the OSM Mapper display, or any other display that I'm aware of. It would be a bit of a challenge to retrieve the data, but might be interesting. For example I'd like to know which bits of the buildings data in London are still based on yahoo rather than bing.

By the way, failing to add a source tag should always be regarded as quite acceptable, in my opinion. It is yet another complication when learning how to contribute. There's a danger that it becomes one complication too many for some people. It also slows things down. Perhaps adding a source tag to a changeset should be supported as a simple drop-drown in Potlatch or JOSM.

OpenStreetMap Blue Posts

more photos on flickr, including a load from Alex (better camera!).

We talked about legalities of printing bing imagery for use while paper mapping. It probably isn't strictly legal, either because of the terms of the agreement between OSM and bing, or because of general rules of copyright for printing (which are mostly ignored for one-off prints) It seems like very much the white end of a shades-of-grey legal situation. In any case printouts of aerial imagery generally don't work well at all so who cares?

With Alex I talked about the Russian OpenStreetMap awesomeness of the Saransk mapping party video. It was a remote mapping party, using bing imagery to map the city quickly as a group. They picked a city where there had been some competing closed-data initiatives recently, including some related to bus routes. At 35 seconds you can see them coordinating the mapping with numbered cake slices, crossed out in a google doc. They've since been developing an in-map clickable javascript cake slice management tool.

Steve Chilton was there (Username Steve8, but is he the real SteveC? You decide) He described a project he's working on at the British Library at the moment, studying the early beginnings of the Ordnance Survey, and exploring the idea that they first developed their map in waves of detail and on an interest first basis rather like OpenStreetMap. Turns out they didn't. It was all a bit more military than that.

I chatted with Andy Allan about his awesome new transport map.

Read about it on his blog post where he describes some of the details and anti-details involved. We talked about that, and about stylesheet generating. A raw mapnik stylesheet is monolithic hulk of XML, so Cascadenik seems like a good idea, but can result in inefficient stylesheets. What about Carto, developed by the TileMill folks? Not sure how it compares. We also talked about hosting mapnik tile rendering. Given the availability of fast SSD drives, you should put the postgres database there, and use slower disks for the tile cache.

...and then we went to node:787474861. Nom nom nom.

The next London Summer Mapping party will hopefully be a mapping party! Details coming soon.

Location: Fitzrovia, London, Greater London, England, W1W 7PR, United Kingdom

Comment from z-dude on 17 April 2011 at 07:17

cool, I shall start using railway=subway_entrance from now on.

Comment from Tom Chance on 17 April 2011 at 08:29

Thanks Harry. OpenStreetBugs might work better if it was a communication channel, i.e. you registered an OSM account and when your bugs are updated/fixed you are notified.

I've started putting the source tag on buildings these days, and I've been retrospectively adding them to buildings I've traced from a combination of OS and Bing. What's interesting about your map is that we already have three different values for OS data, none of which are precisely the value suggested on the wiki page!

Comment from Steve Chilton on 17 April 2011 at 10:07

Here's a thought. It toook the Ordnance Survey longer than they had expected to survey the whole of the country in the 19th century for the first large scale (6") series of maps - over 50 years in fact from start of survey to last map published, despite it being a priority and putting large amounts of staff and resources into the project.

Comment from Harry Wood on 18 April 2011 at 11:13

@alexz. bank tube station is probably the best one for multiple entrances! (think they're all already mapped now)

@Tom Chance. It's a shame OSM Mapper isn't permalinkable. The top 20 source tags look like this (fingers crossed this HTML works):

6536Bing;OS StreetView
1665OS_Opendata_StreetView auto-trace (Tom Chance)
165OS OpenData StreetView
124survey on 2007-09-09 with yahoo and gpx support
104survey on 2007-09-08 with yahoo and gpx assistance
70Yahoo!, survey
64Yahoo, survey

As well as failing to take account of source tags on changeset, you can see there's a lot of semi-colon value separators to deal with too. But yeah quite a lot messy capitalization and other typing variations.

@Steve Chilton. Interesting. Maybe they spent 5 years mapping 45 years bickering about which tags to input into their 19th century map database :-) The process of mapping contour lines sounds like fun. Contour line mapping party??

Login to leave a comment