Either way (London) comes under the county system we’re removing, so I’m for it. Has the wiki changed yet?
I would go along with only locating highways (major & minor).
I presume we’re going to use the town from the post code? Although I’ve seen the discussions about other sources, I haven’t seen anything else that really seems useful. Most of the other stuff seems to be base around electoral wards which (a) change, (b) have names that aren’t really useful and (c) tend to ignore post code boundaries and/or local opinion.
I would think perhaps we should not necessarily “locate” the primary streets. We were also talking about naming “blobs” of roads at the centre of smaller localities to get them on the map. These are the sorts of localities that are probably centred on primary streets. If we haven’t made a rule of locating these streets by post code, the possibility is then able to locate them by “local locality”!
[color=#0040FF]Off-topic. I’m mobile again. 
My ex-wife found a Rover 416 with 6-months MOT, bought it, and drove it up to us on Monday, then caught a coach back to London on Tuesday. I’m sure it was done less for me (she did divorce me!) than for the sake of the kids and their holiday. Cost of car, tax, coach fares etc should end up abou the same as the write-off value of my old car.
I should be able to extend my editing area further through South Wales now - and my god it needs it! :shock: [/color]
Iain, can you talk me through how you get from coordinates to postcode to town?
Glad the other problem is sorted.
David using Tapatalk
I’ve got nothing that gives co-ordinates, I’m afraid. I must admit I’ve been assuming the information would be fairly freely available. I know you can see postcode boundaries on an A to Z, and to some extent they’re shown on the street signs.
Part of the reason I’ve been keen on the “main roads only” idea is the thought that we would still have to manually look up the boundaries: by reducing the number of roads we locate, we reduce the number of boundary crossings we have to identify.
A quick bit of Googling brought me to this site. I don’t know if it’s “free” enough for us to use.
Presumably you could follow the boundary in WME until a highway crossed it. Make sure it’s split at the boundary and located on either side. Once the boundary splits are in place then “select entire street” for the bit in the middle would hopefully select from (not including) one located boundary segment to the other. Then locate that centre section.
Why would that be any better than the source at post 28?
David using Tapatalk
And you still have to get from postcode to locality.
David using Tapatalk
I thought I posted something there, but seem to have lost it whilst serving dinner!
As for the source from post 28, it may well be better. I didn’t actually download that one and look at it. I still consider myself an inexperienced editor, so I’m sticking to the manual WME editing until I gain experience. Until I’m ready to think about trying to become an AM, I haven’t been thinking in terms of coordinate-based sources and mass-editing tools.
As for postcode to locality name, this page on Wikipedia has the names to correspond with the individual postcodes.
Once we get on to the smaller localities, I don’t have anything definitive to suggest, other than local knowledge. For example, I know that, whilst “Rushey Green” is a local ward, nobody refers to it as an area - the name indicates the name of a main local street, so I wouldn’t try to make a “Rushey Green” blob.
A problem with London Postcode Map is that it’s inaccurate - quite severely so in my immediate area with the boundary between SE19 and SE25.
My copy of the Nicholson Greater London street atlas includes postcode boundaries on the map, and as far as I can tell, they are accurate, at least in my neck of the woods. That’s fine for outward codes, ie SE15, CR0. Localities generally share the first digit of the inward code: Shirley (Croydon) is CR0 8xx; Addiscombe is CR0 7xx and CR0 8xx.
It looks as the Code-Point Open data set comes with a full set of complete postcodes and the northing and easting for them. The other data that comes with it will also be useful. Brain’s about to shut-down for the night but some spread-sheet manipulation of the data that’s supplied there could well be a long way to a solution.
Unfortunately, Code-Point Open comes on a disc and takes a month or so to arrive. I’ve requested a copy, and it’ll be useful whatever.
If you’re on about the OS OpenData product, then it appears to be available as a download too…
Code point afaik is not licensed under OpenData, so for the love of god be careful with what you’re using. 
Sent from my GT-I9100P using Tapatalk 2
Code Point might not be, but Code Point Open certainly is, although in order to use it you do need to include an additional attribution statement:
So when someone gets the product please can they post a summary.
I think what we need is either:
A product that gives us precise boundaries of whatever localities we wish to use, even if we have to go via postcodes or some other boundaries to do that. We can then use that either manually or automatically to name roads’ city field (all or just majors etc as we decide).
A product that enables us to locate the centre of a locality and to name all roads’ city field at that point, with all other city fields being blank.
I think the argument was tending to the former but please correct me if I’m wrong.
The Code Point description (which corresponds to what I noticed in a quick peek at the data last night) on the OS site says it contains the following:
Postcode units, eastings, northings, positional quality indicator, NHS® regional health authority code, NHS health authority code, country code, administrative county code, administrative district code and administrative ward code.
The potential point of interest here is the inclusion of district coding as an intermediate level between county and ward - we know the former is too imprecise and the latter too precise, so having some data that defines localities at an accuracy somewhere between the two could be quite handy. I didn’t look at the district coding definitions last night, so I don’t know just how precise an area they define - job for me this evening if no-one else has grabbed a copy of the data by then…
If we just want to generate naming blobs around the centre of each locality, this should be possible using the OS Gazetter data.
I was going to suggest the Gazeeter actually, it might help getting some blobs down.
Sent from my GT-I9100P using Tapatalk 2
No good, for the London area the district coding is accurate only to the borough level, and we’ve already got access to that level of accuracy in the OS Locator dataset. So far then, it’s looking like using the Gazetter dataset is the only option we have for (semi-)automatically generating reliable place names, which brings us back to the name blobs idea…
So, just to be clear, you’re saying there appears to be no available free source that gets us to exact “locality” boundaries, either directly or indirectly? If so, that’s a bummer
Correct. Now, does it matter. We know we can place accurately located locality data centrepoints around the map based on the gazetter dataset, so are we losing anything by not being able to definitively place each and every street into a locality?
I’m really not sure. My brain is hurting today.
David using Tapatalk
No, I don’t think accurate city boundaries are important to Waze. Cities are only used for reports, and reports tend to only appear on main roads.
(However, my OCD disagrees with me; I’m trying to smother it with a pillow)
I dont think it does matter. If the little blobs are there the localities should show on the client and hopefully give “near…” for the traffic reports.
I’ve done an experiment here in Wales, where the map is so bad I can’t make it worse! I’ve named one village by locating the local roads around the A484 (about a dozen), then the next by locating the segments of the A484 that pass through it. If we get a map update before next Friday (!) I’ll see how it works.
As for the postcode boundaries - if I have to sit there referring to an A-Z whilst locating the main roads in SE London, it will be a slow job but worthwhile!