# Latino House Numbering Scheme

I know this discussion was abandoned a long time ago due to lack of agreements and the different ways to identify houses in each country.
Anyway I think we should start with something which may not be perfect but at least help people to come closer to their destination and at the same time offer an easy way to introduce house numbering to OSM
Therefore I would like to make an approach to propose something which attends very well the needs of several countries and comes closer to others which have a more complex numbering scheme.

For those countries which use only or mainly numbers to identify houses and these numbers are more or less distributed on a regular basis over the blocks. E.g street “Hummingbird” has got 20 blocks, each block has 100 numbers, meaning
1st block numbers between (0-99)
2nd block (100-199)
3rd block (200-299)

last block (1900-1999)

We can build a relation which uses the, already existing, intersections nodes to set the start number of each block and the way between the two nodes to set the street-block to be numbered on a sequential basis.

Below is a first draft of such relation
It is possible to:
a) Build a relation to set numbers either both sides or just one side of the street. So as to attend irregularities
b) Break down the relation so as to tag a street in parts in case of irregularities. First 5 blocks with one relations, next block with a different one
c) Extend later the relation to add more complex cases
d) Set postal codes to the left or right side of the street, in case they may differ
e) Set odd/even numbers to each side of the street

The members of the relation has to be ordered either increasing or decreasing way
The nodes have the role to hold the start number on the specified block
The ways have the role via getting assigned the house numbers between two nodes from previous node to next node-1

. . . .

I think this sort of relation would solve the house numbering issue in several countries like South America and at the same time be a good approach for other countries.
Even for those countries which use a combination of numbers and letters, like 20a, 20b, 20c this method will route you very close to the house
We can’t expect OSM routing you exactly into your girlfriend’s bedroom.

The effort of such numbering scheme is minimal compared to others, there is no need to add new elements to the map. The same node can be reused in several relations to number one street and the crossing one. Right and left side can be created in two different relations if needed.
One street can be numbered in more than one relation attending certain irregularities, even a block can be split in the middle if needed.
A block with no houses can be left out of the relation (2 relations are needed in this case)
The relation is easy to extend so as to add other cases

Hi Beerforfree,

I suggest you to post your idea on the tagging mailinglist. (but pay attention the tone might be crude).
Another good way is to open a proposal page to get a plattform for discussions: http://wiki.openstreetmap.org/wiki/Proposed_features

My first impression is that it seem to be to complicated (relations) but I will give it a second view

You didn’t like the girlsfriend’s bedroom part, right? ha ha ha
I can remove this part for sure, here at the other end of the world we may have a different sense of humor to emphasize things

I really appreciate your feedback and hope to get more feedback before submitting a proposal
thanks my friend !!

Why not just the standard scheme for addressing, which is already in common use, and supported by applications etc. See http://wiki.openstreetmap.org/wiki/Key:addr
Yes, with the addr scheme you would have to create extra addr:interpolation ways, but it shouldn’t take long to add a pair on each block.

I think your proposal looks more complicated, and more likely to have errors, as you have to create relations and add the correct nodes to them.

Here an example of a couple of streets in a village. Even more simple than the original proposal, just one way and all the nodes who participate in the relation. Look how simple and flexible it is, you can even left out parts of the street with no houses (boulevard crossings)
In less than a minute we can set numbers to a whole street without having to draw a single node

http://www.openstreetmap.org/?lat=-35.2605101466179&lon=-59.6296316385269&zoom=17

But its got nodes in the middle of junctions with numbers. Unless there is a house with that number in the middle of the junction, that’s just wrong. And you’ve got nodes in the middle of a junction with several different numbers for different streets.

Also, its unclear which direction the numbering is in, and which side is left and right. Relations do not have directions.

IMHO such a relation doesn’t make sense. Still the best is to set a node for each housenumber. Already the existing interpolation is just what it says an approximation. In general houses and land plots tend to be irregular distributed and you’ll have gaps. There are plugins for JOSM to ease entering equal distributed nodes. Or as an alternative only set housenumbers at the corners.

Housenumbers in South America have gaps. Using addr:interpolation you would map housenumbers that don’t exist.

Of course, that would be the best! But you forget that e.g. in Argentina, there live about 40 million people - and you have 10-20 (really) active mappers. Schemes that work in Germany (I use KS here.) are too much work in other countries.

Similar to Thailand where I’m living and mapping: 60 million inhabitants and I’m afraid less than 10 active mappers. The houses get numbered as they are built: e.g. here, house numbers 1, 2 3 in the center.