OpenStreetMap

mwbg has commented on the following diary entries

Post When Comment
Random bus routes 10 months ago

srbrook: yes, I was sort of aware about how to sort, move, reverse (subsets of) the ways in a relation but, as I've pointed out elsewhere, with the case of self-crossing bus routes, to do sorting like this would break (or at least "flatten") the route, making it rather pointless. As I understand it, you're a cycle person and can thus follow what ever subset of a cycle route you like. If you're using buses (or even trains), you don't get a choice: if the bus serves this road before serving that road, in a distance-inefficient way, then that's what it does. People who are not regular bus users are perhaps not aware of this.

You could have a route that follows A->B->C->D->E->C->B->C->E, all in the same "outward" direction. Without ordering, it would just look like A->B->B->C->C->D->E. You wouldn't be able to tell that it had two loops and one reversal in it.

I've even seen some relations that omit the "forward/backward" distinction, making it completely ambiguous, even for a human being.

Random bus routes 10 months ago

Vincent de Phily: I don't know about tools but, until now, the notion that a relation could be unordered had never occurred to me. I'm currently using relations to map bus routes, where the order certainly is significant: it makes a difference whether a bus branches "this" way or "that" way, and whether it does so in the nb/sb direction, and whether it does so on the first/second pass of a self-crossing loop.

Current guidelines for route=bus expect the stops to be kept in_order and either "all before" or "all after" the set of ways.

I don't know about other uses of relations, but for transport routes which are anything other than plain linear, it damn well does make a difference, and certainly is not cosmetic.

Gluing, Landuse 12 months ago

malenki: I was being a bit hypothetical in my original post. The specific problem is that I am mapping bus routes (relations) on existing roads.

For one thing, it's very difficult to get at the road (in JOSM) because it's glued to (in this case, another instance of the same road, and) to two landuse boundaries.

In practically every case, the adjoining landuses are residential, as are the roads going through it/them.

As far as I can see, there are two intepretations, and the gluing that I am seeing doesn't match either:

I suppose my question is: does the road inherit the landuse of those polygons through which it runs, or does it have its own landuse (as yet, undefined).

In the first case, the centre line of the road should form no (glued) part of either landuse, and the two landuses should be merged.

In the second case, the two landuses should be separate, and end at the road margins.

In all the above, I'm assuming that the landuses are identical, have no other different tags.

Complex Bus Routes about 1 year ago

Ordinarily, I'd agree with you: maintaining multiple variants of a bus route is a bit silly.

Thing is, my local bus company has historically used a "route number" (in this case, 44) to refer to what amounts to a whole bus network within the collection of little districts that is WR14. It's all called the 44 and has significant variations on Saturday, Sunday, Schooldays, within individual days, and according to drivers' whims.

These aren't just minor "doesn't serve XXX after 1900" variations: WR14 has no obvious terminus: it is served by what amounts to a large terminal one-way loop. There are at least two major variations of this loop, altering the duration of the route by about 30 minutes. In addition, short-runnings are common.

Think of the District "Line" of London Underground, or the Metropolitan "Line": these aren't "line"s at all, but collections of overlapping routes with a variety of service patterns, routes, and calling points.

There are "A", and "C" variants of the route, but these variants are minor compared to the variants within the route, especially considering the number of short-runnings.

There is also a large extension within Worcester, to "County Hall". Only certain buses go there, and there is no correlation between those that do and what suffix letter they have.

FMR (the operator) are not very good at publishing coherent maps and information as to where their buses go and the buses often (more often than not) run with wrong display blinds (mainly because not everyone can agree where the logical terminus is). Look at the timetable and you'll see what I mean.

I wouldn't be too bothered if a relation didn't have to be a strict linear collection of ways/nodes. As it is, I have to do it in pieces and, given that it's in pieces, I might as well take the opportunity to re-assemble those pieces using super-relations.

mwbg

osmrm about 1 year ago

Yes, I'm familiar with the osmsurround site. It doesn't do super-relations, however. Hence my query about osmrm. The latter is still down, after four months!

Doing an A-road as a relation about 1 year ago

@PinkDuck. I must admit I hadn't thought of the roundabout question.

I have been led to believe that each and every road segment in the UK has no more than one letter-number. There are numerous examples of e.g. where two A-roads go through a single thoroughfare in a small town, but where that thoroughfare has just one of the numbers, with the "other" road disappearing on one side of the town, only to restart on the other side. I've only ever once seen an A-road marked as Axxx/Ayyy on the ground.

So, given the choice, in your example, I'd leave the roundabout unnumbered, with two disjoint segments of road either side. The plotting of a continuous representation of the "Axx" should be done at rendering level.

I think there is very little point in maintaining a relation where that information could be unambiguously derived from the map; it's just one more thing that people have to (but shouldn't) fix if they, for example, add extra bend points to a road's geometry.

I will leave extant relations of this nature alone. I won't delete them, but I won't go out of my way to maintain their continuity should I have cause to, for example, retag a road due to the council renumbering a section of road (such as downgrading the A999 to the B9990).

Bus Routes and variants over 1 year ago

The route in question is FMR's 44 route. Looking at the timetable, there seem to be several short runnings, extensions, loops, opt-outs, runnings-under-different-number. Yes, super-relations seem to be the way to go. However, none of the analysers seem to support them yet.

Oh dear. I wish all this stuff in OSM were defined properly, without a chaotic mess of proposals, committees, ideas, wikis. I don't want to direct policy on OSM; I just want to follow rules and map my area.

FMR are fiddling again. This time, they've invented a new route called "43F". (Who ever heard of a bus suffix "F"?).

I've added a tentative role "school_loop" until someone can tell me how to do it properly. I'm not even sure that the loop is correct: the timetable is not very forthcoming.

This new route also has a one-off extension (not a loop) to serve a (different) school. Should I just make this part of the route, and just accept that the normal daytime route is a short-running of that? Seems a bit silly.

It also has a "request loop"; on request, it deviates from the main road to serve an industrial estate (SDO). Naturally, it doesn't pick-up there. Should I show this loop separately, or just consider the loop as part of the "main" route ? Serving the request loop doesn't omit any stops on the real "main" route.

Anybody know what's happened to Matthew Somerville ? almost 3 years ago

@Mappo
I wasn't aware that more uptodate data (2011) had been released. I went to the thread in question and grabbed the two .xls files.
I noticed that at least four that I know to have been stolen/removed in 2009 are still present on the list. Not very reliable, therefore.

Lonely no-entry signs about 3 years ago

Half my respondents said to do it as a plug; half said to do as a turn restriction relation.
I eventually did it as the latter. Can someone check I did it right, please ?
The coords are 52.1119981 -2.3289669.

Creating nodes from EXIF data in JPEG images about 3 years ago

@vclaw Great. I didn't realise you could do that from something as obvious as File|Open. Ih also hadn't occed to me to use d&d. Thank you again.

Creating nodes from EXIF data in JPEG images about 3 years ago

@Sander17 Yes, of course there are going to be loads of pictures that are wrong/bad/unhelpful. Of course I am going to have to do lots of postprocessing.

Your second paragraph suggests that JOSM already does that. If that's the case, then would you mind showing me how to do it ?

I want to upload a load of pictures *to JOSM (or similar)*, perform the necessary postprocessing there, tidy up all the nodes and then upload that to OSM. If I can already do this, then show me how.

Creating nodes from EXIF data in JPEG images about 3 years ago

@TomH So it would record a location ooh a whole 40cm away from the actual object !
I normally take several shots of a postbox, so that I can correlate it with other objects, post some to flickr, and read the ref data of the CP. I would, of course, see several nodes for one real object, but I think I can work out which is the real one.
Also, even if I stand say 60m from a public house, to get the name, operator, it would still be obvious which PH it is: you are unlikely to get two PHs within 60m !

More postbox nemeses about 3 years ago

Hmm. It seems that it was repeatedly blown-up and stolen. When it was replaced by RM, it fell off again, so a local Parish Councilman took it down himself. RM aren't going to replace it as there's another one a few 100m away. dracos isn't updating the list of missing/stolen PBs very well currently.

What's the point of all the TIGER tags? about 3 years ago

What on Earth is TIGER ? I've seen loads of these and thought these comments would answer that, but all the respondents all seem to know what it means !

MWBG

No Motor Vehicles Except For Access over 3 years ago

Thank you, Tordanik and Sanderd17.
I will indeed use motor_vehicle=destination

In this particular street, there is nowhere for bicycles to go (the only exits are cycle-prohibited footways). This doesn't stop a persun physically picking up a bicycle though as (in the UK), it ceases to be a vehicle. This is the same if you just wheel a bicycle (but I know this isn't true in other countries)

Anyway, irrespective of what a Police persun would be likely to do, I would like the tags to match the actual road signs. In this case, it says "No MOTOR Vehicles"; there is a "No Vehicles" sign (which no-one ever understands because it's just a white disk with a red border). I've never seen one of the latter qualified with "Except for Access", but such a Traffic Regulation Order could exist.

Disappearing Post Box almost 4 years ago

Thank you to all those who pointed out that GSV do the odd infilling, even of single frames. Also, thank you to the chap who pointed out the other inconsistencies (such as the fact that it wasn't raining).
Might be an idea to find out where the postbox was before it was moved to its current location, however.

Place names masquerading as road names almost 4 years ago

@netman55.
I hadn't thought of looking on Estate Agents' sites but I did so and got the same results as what you did.
I also went to the RM postcode lookup; it's very non-forgiving when you want to turn an address into a PC.
The format for Pin's Green is "xx Pins Green" as the building name then "Worcester Road" etc as normal. Thus, I would treat that one as a mews or other dependent thoroughfare (i.e. PG is subordinate to the main road)
The format for Lower Interfields seems to put LI as superordinate to the main road, i.e. "Leigh Sinton Road,.. Lower Interfields".

So, LI is a hameau; PG is a mews.

Now, what to do about "Waterloo Close".

I just wish they'd use different name plates for "places" versus "roads".