Hmm, good point. Oh well, at least I got to play around with sqorl a little.

Using the metacarta up-to-date bookmarklet the Österreich label disappears. That may mean it's already been fixed or the official version of Mapnik might be doing something strange that metacarta isn't.

OK, so if I do manage to visit this part of the country then I need to take a look at whether the trunk routes are suitable for cyclists and pedestrians. I'm not entirely convinced that's the only reason for bad routing, but it may be one reason.

I have two screenshots from Josm to give an idea of where they all are but they only show as unrecognised node types. Current oil wells and Disused oil wells

The DECC were keen to state that they hold no responsibility for any errors or omissions, so they may be aware that this dataset is less than perfect and aren't willing to say so directly.

There is a preliminary proposal for using relation=site in that situation, but school is also valid for an area so you can add the boundary of the school and have area=yes;amenity=school for the area.

Buildings within that area don't need amenity=school, and would just have building=yes. A playground within the school would probably be area=yes;leisure=playground;access=private.

If the buildings have seperate addresses (as in a large campus) then you would tag the buildings with addr:street=* but if they all have the same address then the area would get the addr:street=* tag.

The wiki shows which amenity tags are valid for areas or nodes or both. Schools and kindergarten should be acceptable as areas. As for shops, they come under the shop key and vary as to whether the tags are valid for areas or separate nodes. So sometimes you will be fine leaving the areas tagged, and not have to make changes.

I'd try and avoid having duplicate data if possible. So where you are adding new nodes because of this restriction, remove the tags from the area.

I've been following the talk list, but not seen any real outcome. Based upon the wiki page for key:place, St Davids should be demoted to village (population below 10,000) but then it would be technically wrong and wouldn't get rendered until too high a level zoom.

I understand about the whole not tagging for the renderer, but maybe a way of telling Mapnik which places are more important to display is better than the technicalities of what is and isn't a city.

I've been following the talk list, but not seen any real outcome. Based upon the wiki page for key:place, St Davids should be demoted to village (population below 10,000) but then it would be technically wrong and wouldn't get rendered until too high a level zoom.

I understand about the whole not tagging for the renderer, but maybe a way of telling Mapnik which places are more important to display is better than the technicalities of what is and isn't a city.

OK, quick rough and ready count from an xapi export of railway=*:

3334 stations (people have checked that all mainline stations are included, the extra 800 or so stations here will include stations which are tagged multiple times and non network-rail owned stations

7938 meters of line. Hmm, something wrong there, let me see...

highway length sums (metres):
rail 17212699m
abandoned 4333055m
preserved 579200m
subway 437247m
disused 316685m
tram 191097m
narrow_gauge 141371m
light_rail 108768m
station 31080m
construction 20172m
platform 19184m
freight 8031m
miniature 5698m
monorail 3416m
tunnel 3325m
service 2463m
narrow_gauge; abandoned 1914m
incline 1851m
depot 1339m
viaduct 931m
yard 918m
narrow guage 842m
siding 755m
Bristol Harbour Railway 750m
PRT 730m
Platform 310m
level_crossing 206m
signal_box 145m
abandoned; disused 106m
fake 104m
suspended 69m
proposed 61m
cliff_railway 51m
crossing 26m
(type value here) 24m
TOTAL 23424644m

which works out as 10,700 miles of track which I assume is network-rail owned. However, this doesn't take into account the infrastructure or multiple lines running alongside each other which either aren't on the map or are tagged with lines=*.

bridges = 2687
tunnels = 506

So 3100ish tunnels and bridges, but that's across the entire rail network and not just network-rail owned infrastructure. This figure seems very low.

I'd considered that, however whilst I can derive those figures from the osm xapi export I have it wouldn't be too accurate. I'll have a go at pulling these figures later, but:

Stations can be tagged multiple times in OSM because of platforms not ajoining each other
Track and infrastructure isn't too meaningful, infrastructure would include highway=service lanes around the railway which aren't tagged as railway=*, tracks aren't all owned by Network Rail, etc.
Bridges and tunnels wouldn't account for bridges going over railways, and we don't know if that is counted, plus tunnel=yes would cover much of the London underground system.

The grey areas come from Mapnik rendering, and I haven't actually looked into where the data for them comes from.

Mapnik seems to not render things exactly as I request sometimes, which is why I haven't got a key made. This is high on my priorities.

I had a go at adding a permalink function, and it seemed more complicated than it should be. The Openlayers permalink function is easily added, but you then need to get the php to use it when clicked on and that was conflicting with existing code. Again, it's high on my priorities.

There are theories about how you make GPS work using flat earth theory, but they involve big radio masts. Yeah right...

We've been discussing that elsewhere, and yes - that is what I am aiming to work on today. However, the importing of both waterways and railways into my postgis database is currently failing and so it may take longer than it should.

Hill shading is going to use a lot more cpu and memory than the little rendering server I'm using has. It's struggling with the addition of waterways. I'm aiming to come back to this tomorrow to at least add in the Thames, for rail maps I'd consider hill shading overkill but you've got me interested now so I may try and add it.

What did you have in mind? If it's easily retrieved using xapi and adds context then I don't see why not.

Oh, and I didn't add stops for the X6 - on account of not knowing which ones it should stop at. And the X6 also runs in the opposite direction, which I haven't added but perhaps could do most of from memory.

OK then, I've added the X6 route in as much detail as possible. There's a couple of places where the roads look a little odd (the guided bus route Manchester Road and the Leeds inner ring road joining Park Lane).

osm.org/browse/relation/58840

The 363 I have left unmapped because I only have part of the route, and the 72 I have left unmapped because gps signals were terribly weak in Leeds city center so I can't do it accurately.

I've been looking for something like this, between these two I can now render the Lycian Way relation nicely to make it clearer what's missing.

The main map and cycle map are typically rendered once a week, so it may take anything up to a week to appear. The osmarender version is rendered continuously and will typically be updated within an hour.

Aw shucks, does that mean I missed barcampsheffield then? For some reason I'm always a week out in trying to remember when the local barcamps are :-(