“Was there ever a time when relations were not ordered?”

Yes. Relation ordering was introduced with API 0.6. Relations were not ordered in API 0.5.

Relation ordering is still optional and most tools don’t make use of it. By all means order relations if it makes you happy, but with rare exceptions (e.g. buses that pass over the same section twice), ordering can and should be derived algorithmically.

No, Chris is right. We could have imported all of the OS vector data automatically and we chose not to. We could have imported all of NAPTaN automatically; we tried a few bits and chose not to.

The fact that OSM was a (partial) contributor to achieving the OS OpenData release (though the Guardian’s Free Our Data campaign and the Cabinet Office’s open data initiative were more significant) is hardly relevant. OS OpenData still did (and does) contain a lot of stuff that isn’t in OSM and that we could have chosen to import had we wanted to.

Right, and it’s a valid question to ask what methodology makes the best free map of the world; and, in particular, whether the community that (everyone agrees) is required to update the map grows best when presented with a ‘fait accompli’ imported map, or when it’s able to evolve the map at its own pace.

You are welcome to draw your own conclusions on that score, but it’s worth noting that the two best-mapped countries in OSM are probably Germany and the UK, neither of which has any significant imports.

Only if you add meaning with surface/tracktype tags. Tagging, say, the 7Stanes and Coed-y-Brenin trails with highway=cycleway and nothing else would be misleading - there would be nothing to differentiate it from the common-or-garden urban cycleway or Sustrans rail trail. Happily, they do indeed appear to have surface/tracktype tags.

That’s the point here. highway=cycleway has implicit meaning; look at 90% of cycleway usage in OSM and it means more or less the same thing. But if you want to overrule that implicit meaning, you can do so.

highway=path doesn’t have implicit meaning, other than “here is a path of indeterminate construction”. Unless you supply the meaning with both access and surface/tracktype tags, it’s pretty much useless.

That’s absolutely it. highway=cycleway tells you it goes “whoosh” and you can cycle on it; it works in E&W and elsewhere. highway=path and an access tag doesn’t tell you it goes whoosh in any country.

Superb. Really hope this gets adopted widely.

@Tom: and yet people post to the diaries, even though they’re not as fully featured as Wordpress (though they’re probably not that far off Blogger…), because it’s where OSM people gather. People use the messaging system, even though it’s not as fully featured as e-mail. Groups are just the same: we’re not trying to out-feature Facebook, we’re providing a service for our users. You try suggesting to a bunch of mappers that we all use, ooh, G+ and see where it gets you. osmf-talk refers ;)

What really excites me about groups is that they actually reflect how OSM works at its best.

Everyone has their own focus for OSM. Sometimes it’s locality: they want their village, town, city to be mapped well. In a few places we provide tools to help this happen - there are a few city/county-scope mailing lists at lists.osm.org, and other places have their own Google groups or whatever. But that’s incredibly patchy (most places don’t have one), unscalable (if hosted at osm.org), and requires the sort of techie mindset that “does” mailing lists.

Often, though, it’s thematic rather than local. My preoccupations are mapping the UK’s National Cycle Network and its waterways. It would be terrific if there were somewhere to talk with other people of similar interest, so we could feed off each other and get the job done better, quicker.

Right now there’s nowhere to do that. You can do it on a general-purpose mailing list (in this case, talk-gb), but you’ll get howled down by the marauding wolves who have no interest in the NCN/waterways but have an opinion about everything. You can ask Mike to set up a mailing list at osm.org, or use the aberration that is the new Google Groups, but that’s a barrier for something that should just be “hey, we’re having a conversation”.

I see your point, Alex, that “today OpenStreetMap enthusiasts gather in spaces on mailing lists, Meetup.com, Twitter, Facebook, forums, and Google Groups”… but honestly, I don’t think they do. Not over here, anyway. Super-connected guys in metropolitan areas are doing so, I guess, but that fixes London and NYC and SF - not the rest of the world. Even here, where the stereotype is that all our mappers go to the pub, we have regular pubmeets in London, Birmingham, maybe Edinburgh and Nottingham, and that’s it: a tiny fraction of the community. Most of the names I meet in the edit history are untraceable through our existing social forums; they really are just working in isolation.

I also really see this as a way to prioritise “local knowledge” at the expense of “m4pp1ng sk1llz”. Right now it’s easy to get the impression that OSM’s most valued contributors are those who’ve made the most wiki tag proposals and who juggle an armoury of 537 JOSM plugins for their mapping. You know and I know these aren’t the guys who make a difference, but it’s not how it looks - and I think we’re already seeing the first signs of the “Wikipedia effect”, where the top 5% exerts such pressure to conform with their arbitrary rules that newcomers are warned off. Baking thematic, interest-based social features into the core of osm.org is a really good way to demonstrate what we care about.

Yesh, Mr Connery.

Certainly open and connected. But the surface hasn’t been marked in OSM so it’s not clear how passable they are. Have a go and report back so the surface can be recorded on the map!

No, no-one owns it. “Copyright” is the right to copy. If you have independently come up with the same fact, you haven’t copied it from (say) Google. But that doesn’t mean that you could just copy the fact from Google without coming up with it yourself.

All of this is very jurisdiction-specific but you may want to look at the ‘Case law’ and ‘Statute law’ pages on the wiki.

In many countries, rights can exist in a collection or selection of facts/locations. In addition, many websites (such as Google Maps) have Terms of Service that forbid you from copying information, even if copyright would permit it.

OSM’s attitude is, and always has been, to play safe by not copying. We do not put our entire project’s viability at risk by testing out legal theories.

For more details, see about 987,000 posts on the legal-talk mailing list in the past 9 years, a bunch of wiki pages, and so on and so forth. :)

Looks great. LAB colours are really good for colour differences, and simple to code: see P2 code at https://github.com/systemed/potlatch2/blob/tracer/net/systemeD/potlatch2/tools/TracerPoint.as .

The Potlatch 2 branch was working really, really well (if I say so myself :) ). I actually traced large parts of SW England and NW Wales using it when the licence changeover was imminent; two major contributors in the area hadn’t agreed to the new licence, and auto-tracing from OS StreetView imagery provided an easy way to save the roads.

The main reason it was never deployed is that I hadn’t figured out a UI for defining the limits of the tracing operation. The current viewport was usually too small, the ‘sum total of loaded imagery’ too large (and too random).

I may resuscitate it one day, but until then, you can see the rest of the code at https://github.com/systemed/potlatch2/blob/tracer/net/systemeD/potlatch2/tools/Tracer.as . Effectively the workflow is:

1. Flood-fill (startPixelIndex/runPixelIndex)
2. Thin to a skeleton (runThin/thin)
3. Mark junctions (makeWaysFromSkeleton)
4. Remove unnecessary junctions (makeWaysFromSkeleton -> thinJunctions)
5. Create pixel chains, i.e. polylines at pixel resolution (createPixelChains)
6. Delete dupes (createPixelChains)
7. Run Douglas-Peucker over each one to make sensible polylines (createPixelChains)
8. Convert to OSM nodes and ways (createPixelChains)

There’s a never-deployed Potlatch 2 branch which uses its own code rather than the MagicShop service: https://github.com/systemed/potlatch2/tree/tracer

And Mapbox have been working on similar things: http://www.mapbox.com/blog/user-friendly-guided-feature-extraction/

“it” is the neuter third-person singular pronoun in English. Steve Coast explains “asl” in this photo. “ngo” is a less common form of “ngggggggggggggggggggggh”, which is an expression of frustration at the latest buffoonery inflicted in the wonderful world of OSM. HTH. (Hope this helps.)

Standard practice in the UK is indeed to leave roundabouts un-ref’d.

@PinkDuck: indeed most current tools do make it easier to show a relation than the set of ways with a certain tag; but as a general rule, since mappers are our most precious resource, we optimise for the mapper. That means not adding duplicate information where there’s no need to do so. :)

Yup, route relations have no purpose in Great Britain. In other countries, where roads can share a number, yes, they are useful. But here they’re simply a hindrance - yet more one thing to confuse the newbie mapper.

Without seeing the trace it’s difficult to know, but the usual reason is that you don’t have timestamps in the file.

OVER 9000

Is there a licence posted anywhere for the Sectional Appendix?