[Talk-GB] OSM Analysis updated with May 2014 OS Locator data

Dan S danstowell+osm at gmail.com
Thu May 15 07:49:19 UTC 2014


2014-05-15 8:27 GMT+01:00 Steven Horner <steven at stevenhorner.com>:
> Thank you all for the advice, although it may have confused me all the more with different suggestions.
>
> Personally I like Marc's suggestion of using the 2 street names separated by a hyphen. This allows both names to be rendered.

Seems OK to me (though I would have chosen slash, but whatever), and
we can fondly hope that renderers will one day detect the "name:left"
and "name:right" tags and render those cleverly when found.

> Then identifying each street with left and right tags. How do you chose which is which if the road runs East to West?

Use the direction of the way, i.e. the direction in which the OSM
object was drawn.

> I'm amazed this doesn't crop up constantly, any old terraced streets with a road separating them would have the issue. I can think of about a dozen streets within 1 mile of me where this is the case.
>
> I will do some more investigation and look at several different mapped areas to see how they have been tagged, doesn't sound like there is a definitive answer.

When I sometimes encountered it, I "solved" the problem by putting the
different streetnames on the building addresses, and ignoring the
issue on the way itself. I wasn't aware of "name:left" etc!

Dan


>> Let me first introduce myself, I'm a Belgian mapper that has been lurking for a few months on this mailing list. The reason is that I want to learn how other communities work and which problems they have and how they solve them.
>>
>> Now back to the topic: in Belgium it's quite common to have streets with two names, at least when they are on the border of two villages. The Belgian community decided to map this as follows:
>> name =  name1 -  name2
>> name:left = name1
>> name:right = name2
>>
>> An example: http://www.openstreetmap.org/way/207455046
>>
>> What are your thoughts about this ?
>>
>> regards
>>
>> m
>>
>>
>> On Wed, May 14, 2014 at 1:07 PM, SK53 <sk53.osm at gmail.com> wrote:
>>>
>>> There are at least two major streets in the middle of Nottingham like this: logically the street does not have a name, the sides of the street have names:
>>>
>>> North of the Council House, the S side is Smith Row, the N side is Long Row
>>> South of the Council House, the S side is Poultry, the N side Cheapside (originally Rotten Row)
>>>
>>> These names originate as locations in the market square, as can be seen by other survivals such as Beastmarket Hill. Where the square is now an open plaza the name of the rows of buildings have been transferred to the thoroughfare. The addresses on Cheapside are even more complex because the shops also have entrances in Exchange Arcade and are let as units of this shopping arcade. The Austin Reed shop appears to have at least 4 addresses from the Royal Mail, OS, Nottingham council & Austin Reed website: all in all a mess.
>>>
>>> Other places where this occurs include: Sherwin Road/Castle Boulevard, where the W end of Sherwin Road has houses with Castle Boulevard addresses on the S side. In this case I resolved it by tagging the footpath with the Caste Boulevard name. This discrepancy arose because the two roads were merged when the roundabout was built in the 1920s.
>>>
>>> I recently noticed a case where the Land Registry data for a small new build terrace had been resolved by using the name of the terrace as a building name. Fail. In some towns (Bangor, N. Wales, comes to mind) many houses were built as named terraces with numbers within the terrace. Although Bangor has been relatively recently house-numbered a simple inspection of addresses painted on rubbish bins suggests that the original addresses are still in use.
>>>
>>> Broadly speaking we should try and do this better than the OS Open Data because it does happen fairly frequently. name:left and name:right can be used even if no-one consumes them at present. It is useful to try and map addresses in such cases, and these are the one case where I am happy to use the associatedStreet relation. This at least enables the correct grouping of entities for the 'street'.
>>>
>>> Perhaps the challenge is twofold:
>>>
>>> Persuading people that streets with addresses might not be named. (The Royal Mail seems generally to adopt a Procrustean solution to force everything to fit PAF).
>>> Working out how to consume such data (mainly for rendering).
>>>
>>> Jerry
>>>
>>>
>>>
>>>
>>> On 14 May 2014 10:07, Richard Mann <richard.mann.westoxford at gmail.com> wrote:
>>>>
>>>> There's one like that in Oxford (for about 30 metres) - street addresses different on the two sides. For the moment it has name="St Clements Street", alt_name="London Place", and a separate footway with name="London Place" (plus a name:note).
>>>>
>>>> So my suggestion - draw separate footways, and give them names. Use name/alt_name on the road, or name = "one name / other name" if both seem equally valid.
>>>>
>>>> Richard
>>>>
>>>>
>>>> On Wed, May 14, 2014 at 9:30 AM, Steven Horner <steven at stevenhorner.com> wrote:
>>>>>
>>>>> Hello,
>>>>>
>>>>> It's interesting and highlights a few problems local to me, some I had buried my head in the sand temporarily because I don't know how to fix them correctly. My biggest problem when tagging roads is what to name a road when either side of the road is a different street. For instance the analysis highlights "Myrtle Grove" as missing here: http://www.itoworld.com/product/data/osm_analysis/map_browser?bbox=415474,536751,415809,537148&referrer=area
>>>>>
>>>>> Myrtle grove is the South side of the road labeled Chestnut Grove and continues around to where the Road is labeled Elm Gardens. Almost all of the streets in the estate are like this, where it is very misleading because opposite sides of the road is a different named street. How should this be mapped, I have steered clear of fixing it because I couldn't find any guidance on how it should be labeled and technically is it even wrong. The actual building footprints I have added the correct addresses to.
>>>>>
>>>>> I use various OS products in my day job and interestingly OSM labels the streets exactly the same as Vectormap Local does, anyone looking at either OS or OSM maps would not be able to find Myrtle Grove. Another street where I have always though was labeled wrong in the village is Roddymoor Road, there is no street sign and I have near heard anyone refer to it as this. The street on part of this road is not labeled (buildings are) it is East Terrace and that's how anyone describing it or looking at signs would describe it. Again OS do this the same which is probably why OSM has it tagged like this.
>>>>>
>>>>> All of this highlights that while OS Locator may have a difference and is fantastic for finding potential problems, changing it so OS Locator comparisons are 100% may not be the correct solution?
>>>>>
>>>>> Any help appreciated and apologies if I should ask in a different list, surely this is an incredibly common problem that I have somehow missed the obvious solution to.
>>>>>
>>>>> regards,
>>>>> Steven
>>>>>
>>>>>
>>>>> On Tue, May 13, 2014 at 2:24 PM, Shaun McDonald <shaun at shaunmcdonald.me.uk> wrote:
>>>>>>
>>>>>> ITO’s OSM Analysis has been updated with the latest OS Locator data. Most places have dropped out of the 100% completeness compared to OS Locator. There’s now 18 places which have less than 95% completeness.
>>>>>>
>>>>>> http://www.itoworld.com/product/data/osm_analysis/main
>>>>>>
>>>>>> Shaun McDonald
>>>>>> Developer
>>>>>> ITO World
>>>>>> _______________________________________________



More information about the Talk-GB mailing list