Recent diary entries
This is an age old-debate: How should one marry up a landuse boundary and a road? Should, with the convenience of Bing Maps, we try and map the land use up to the edge the road on the image - or just connect it up with the road.
Initially when I started mapping I took the former approach, and kept things split. It wasn't until I had noticed somebody had mapped the pavement (or sidewalk for you yanks) along side a road in one village that I started to realise the error of my ways.
Landuse - is just that. The designation of the land - often this is split up by highways (roads and rivers etc). The issue we have is that the line on the map editor is one dimensional - there is no width information, so when we zoom in, the land use edge appears to end in the middle of the road. This is an easy mistake to make as it is what we see but we have to remember when the map is rendered, the road has a width to it and it is entirely sensible for the road to be the dividing section of different landuses.
The alternative approach would be to draw roads as areas themselves, defining the landuse as highway. This of course is a silly idea.
The messy bit arrives when there is another border for the landuse and road - in the form of a hedge or fence. This then gets confusing - where do you stick the hedge? If you stick it in the correct place then usually it's drawn over by the road when it renders. You can't join it on the highway node though, as there are usually a barrier on both sides of the road - and the highway is one-dimensional. Conundrum.
Whilst highways can already have the width defined, I don't think it's something that is really supported or widely used. I think perhaps in addition to this we need to think about having attributes that relate to the edging of roads - this could quickly get complicated though - what about where only one direction has a pavement? What if it has a pavement and a fence?
A while ago I completed the roads for Peasedown St. John in Somerset; put in the key buildings and thought "done".
This was largely done with GPS and a fine day. I was surprised at how inaccurate Google Maps was in comparison, with extra roads left, right and centre.
Then the Bing images were released and I found myself adjusting things and tracing over bits I wasn't able to (easily) map before. One spare lunch hour lead to another and I found myself chipping away at the whole village.
And today, I now declare the village buildings largely done. They're rough and approximate but a good starting place for somebody else to refine.
Comparing the original village (Centre & NW) to the newer homes (E/SE) really shows the difference in space between homes, garden sizes and the "just cram it in" approach modern houses get.
I've done some work on getting the land uses tied up, but there is room for improvement. There is also some clarity needed on paths.
With all this extra information on the map I feel that things are getting cluttered - and probably pointlessly so. Although it's an achievement to get homes on the map, it's generally useless for most map users. I feel OSM needs to take a layer-based approach to rendering, with streets and roads, and then optionally over-lay pedestrian, buildings, etc.
In all, my experience of OSM is similar to other community projects - there will always be dicks throwing their self-imposed rule on everybody else and declaring themselves guru. You either have to humour them or leave the project as they have far more time and persistence than the casual masses. I generally try and find small corners of the map which is ass-hole free.
So I've put a bit of effort in recently to trying to put in all the residential areas around a city. I did it, and only part of it seemed to get uploaded, which wicked me off as I'm generally fairly observant for things such as failures.
Still, the missing bits I did again. They appeared nicely up on the Osmarender layer. Until today. A large number of them have gone. Data and all.
Screw this, I can't be arsed putting time and effort into a system where there is no method of commenting on changes, and rollback versions. If work I put in is simply going to get deleted without comment, then why should I put my own time into this?
Not "the" bug, but another one.
I've noticed that the Osmarender layer on the main page seems to have incomplete loops. e.g. For large round-a-bout style roads or squares.
Compare Queen Square and The Circus on this map:
Saw the article on the BBC News site, so thought I'd check out my area. Not a lot there. So I put a copy of BSGPSPDA on my phone and had a bit of a travel through the non-existent streets. JOSM is easy enough too and runs fine on Linux. Although I was caught out when I didn't close my BSGPAPDA log file correctly before exiting. A quick script easily converts the *.jny CSV files to a structured *.gpx file.
The whole thing seems like a nice idea, but I'm not sure how far it will get. There is no trace of a stream that runs through the valley near my home - and there is no way I'm trudging through it.
I've been doing a lot of work on address data and positioning with my job, so this sort of tool may be handy for a graphical display. But when Google has a far more accurate map, why use this? I couldn't expect anybody to rely on this as there are still so many major things missing.
Keep up the excellent quality of work, I'll try and add bits when I remember. It's quite an interesting thing to do for a while.