[Talk-gb-midanglia] Cambridge projects: houses and shops

[Mark Williamson]

> So the question arises, is there anything left to do other than minor
> tweaks and corrections (which are still alarmingly common!)?
>
> Well, one thing some of the German mappers have started doing is house
> numbering. They've developed a scheme for recording house numbers along
> streets so it is possible to home in on a particular number (I guess
> there's some interpolation involved too, I don't think every number is
> plotted, but I haven't looked at the scheme in detail).

I *think* they do have to plot the position of every number, if it is to be 
rendered on the map.  They have a means of saying "this way includes all the 
even numbers between these two" but you seemed to be expected to plot an 
additional node in the way for each house number.

Their end-goal (according to a remark on the wiki page which may well not be 
an official position) seemed to be to tag individually traced buildings with 
house numbers.

> I notice someone has had a perfunctory go at this locally in the north
> Chesterton area - see the grey blobs with numbers in osmarender:
> http://www.openstreetmap.org/?lat=52.21782&lon=0.13387&zoom=17&layers=0B00F
>TF.

Yup, that was me.  My intention is to do more of the numbers both in this area 
and in the centre but I started off with a small example so that I could see 
how the tagging and rendering works.  I've just not got as far as extending 
the coverage (yet).

>(Also, added individual houses, as buildings -
> http://www.openstreetmap.org/?lat=52.21779&lon=0.13307&zoom=17&layers=B000F
>TF - what do people think about this? Personally I think it hides the wood
> among the trees, and that houses should be at least tagged
> differently so they don't get rendered by default; but that if this is
> to be done, it should be done systematically across the city, obviously
> over time, not just pecked at in one locality).

This was also me.  It was a fairly time consuming session over the Yahoo 
imagery so I haven't gotten around to doing more yet.  I fully expect that 
more will come, especially on these long winter nights.  Most of the buildings 
I've so far done are terrace blocks.

I think it's data worth having since it's fairly easy (though time consuming) 
to derive and can help with orientation.  I certainly see the argument for 
tagging them differently.  Right now, I think if I changed the tags to say 
building=house then they would still be rendered (I don't *think* Mapnik cares 
about the value of the building= tag).  But that would leave the way open in 
future to allow housing buildings to be rendered differently e.g. only at very 
high zoom levels where they won't be distracting to somebody getting a general 
street overview.

Aside from the desire to get all feasible data from our sources into a 
machine-readable form, I think that having housing locations is useful for a 
number of reasons:

1) giving directions to a particular building
2) navigating the various local footways / alleys / cycleways is sometimes a 
challenge.  Particularly around terraces these are poorly signposted and 
difficult to distinguish from private passageways / driveways.  Having housing 
buildings rendered at high zoom makes the positions of these much less 
ambiguous (e.g. do I take the passageway *through* this terraceblock, or is it 
at the end?  Is this path between terraces a driveway or an access road?).
3) It easily enables further, unforeseen computer processing (e.g. calculating 
random statistics on housing in Cambridge)
4) (arguably) shows what the house numbers refer to

> House numbering will require detailed re-surveying, but (a) it is a
> project which is local for many of us, so doesn't involve long journeys
> to seek virgin territory, and (b) there is also the potential to cross
> check what is already mapped.

> Is this of interest to anyone?


Yep, I think that's sensible.  I can potentially get some house numbers almost 
every time I go outside.  House number surveying *is* a bit more tedious, not 
least because trying to use photo-mapping to capture house numbers is liable 
to get me arrested ;-)  But I reckon it's doable and it's all data worth 
having.

> Secondly, while I gathered supermarkets, post offices, pharmacies and
> some convenience shops when I originally surveyed Cambridge, I didn't do
> all shops. The city centre would be easy to do I think, but there's
> pockets of shops all over the place - for example, there's a parade of
> half a dozen I didn't know about before on Akeman Street, here
> http://www.openstreetmap.org/?lat=52.21943&lon=0.11543&zoom=17&layers=B000F
>TF
>

I've also been having a crack at this, from time to time.  I mapped all of the 
shops on King's Parade and some around the market.  I also did a load of 
stores on the retail park on Newmarket Road.

Like the house numbers and the building outlines, I think this is data we can 
derive locally, bit-by-bit that will make a big difference to the amount of 
machine-readable information we have available to do cool things with.

> Again, a local project. We could add a list of places we know about to
> the wiki and then survey them to get the detail. I think I'd not worry
> too much about what is put for shop=... as long as it something
> comprehensible - we can gather them all together later and rationalise
> them.
>
> I will start such a list on the Cambridge page.

Sounds good!  Thanks for the information and for taking the initiative.

Cheers,
Mark