[OSM-talk-ie] Talk-ie Digest, Vol 74, Issue 10

[moltonel 3x Combo]

On 19/07/2015, Nick Burrett <nick at sqrt.co.uk> wrote:
> On 17 July 2015 at 23:08, moltonel 3x Combo <moltonel at gmail.com> wrote:
>> On 17/07/2015, Colm Moore <colmmoore72 at hotmail.com> wrote:
>>> Adding individual Eircodes shouldn't be a problem. Adding the whole
>>> database
>>> is another matter. Facts can't be copyrighted, databases can.
>>
>> And yet there's little value in mapping just a fraction of the
>> Eircodes. Before starting the job, we need to make sure that we're
>> leagally allowed to finish it. If we can't have all the Eircodes in
>> OSM, we should have none.
>
> If it helps somebody locate a house, then it is of value: it is just
> an alternative way of referencing a place.

We're no stranger to having partial data in OSM; we know that
"partially usefull" is still "usefull", even for things that need a
certain completeness level before people really start using it.

What I'm talking about is the possibility of not being legally allowed
to have all the Eircodes in OSM. Being "not complete yet" is very
different from "can't legally go above X% completeness", and if the
second case was true we should IMHO not have the data at all, because
it'd be very different from OSM's usual temporary incompleteness. That
said, I'm pretty sure this FUD will disappear once we get a clear
statement from the PMLH.

> The eircodes are of particular value where a house has no name/number
> at all (which many do where I live) -- how am I supposed to find them
> on OSM?

I suggest using openpostcode instead. For the same number of
characters, you get:
* A pointer to all the houses that aren't on Eircode yet (just browse
finder.eircode.ie they really aren't that rare)
* A pointer to non-postal addresses, like a GAA field, a farm, a music
festival, a beach, a terrain for sale, etc
* A "dataset" that is immediately complete, without having to import
anything to OSM, GM, OSi, TomTom, etc
* No database needed, so it can work without a network connection
* No license needed, so you don't have to pay 5000€ to use it at business scale
* An optional checksum, so that a typo won't direct you 20km astray
* A more easily remembered value, because nearby addresses share the
same first characters.
* The option to give a less precise location, in case you didn't want
to give your exact home address when searching for social services in
your area.

Do not use Eircode. It has zero advantages.

>> I strive to map impartially too, wether I like the place or not. But
>> postcode are not the same thing, because they're not physical. We
>> always think twice before adding non-physical data to OSM. And in the
>> particular case of Eircode:
>> * It's a list of IDs (useless on its own) created and curated by a
>> private 3rd-party
>> * It has an impressive list of technical flaws which make it a
>> non-starter for many usecases
>
> Given it's a list of unique codes for all addresses across the
> country,

No. Due to the way it's designed, they'll forever miss some, and will
be playing catchup^W^Wable to charge for updates. Imagine the fun of
not having an Eircode for your property yet and needing something from
an institution that *requires* an Eircode.

Food for thought: Eircode is fundamentally different from every other
postcode in the world because it requires such a level of constant
updating. Most postcode systems define areas rather than individual
points, so that they remain pretty or completely static. Oh, and Yemen
uses an openpostcode-like system :)

> you can use it as a primary key in a table and populate that
> table with other data like a internally created routing code, loc8,
> openpostcode or whatever else you like.

I don't see any usecase for storing a PostCodeSystemA ->
PostCodeSystemB map. If somebody gives you his address including a
postcode, you store that postcode and maybe the corresponding lat/lon
coordinates. If you wish (for display purposes ?) you can turn those
coordinates back into OpenPostcode or generic Loc8 (but not back to
Eircode unless the coordinates are spot on).

Also, if you're deciding on a primary key for your database, you
should make sure that the key has a value for all the objects you
might want to put in your database. Eircode is the wrong choice for a
primary key: by design it only points at certain type of objects, and
even in its chosen field it might not be up to date and not have a
value yet.


>  It's quite easy to
> rationalise the data into something routable, usable or relevant to
> ones needs.  The needs can vary per organisation.  I personally don't
> need a sequencing system to find a house whereas the companies that do
> can invent their own to suit their needs or translate to a sequencing
> system that has already been designed to suite their needs.

Sorry, I don't understand what you're getting at here.


>> * It's brand-new, and we don't know how much real-world usage it'll get
>> * Mirroring it in OSM is a huge amount of work and a lot of data
>> (which will qualify as bloat if it isn't useful)
>
> Putting house names and numbers into OSM is also a huge amount of
> work, but that hasn't stopped efforts to do that.

Housenames/numbers are a sane system. They have been part of the
addressing system "forever" and people actually use them. I might
change my tune (to a very sad one) if a significant number of people
actually use Eircode regularly. I wouldn't hesitate to map an Eircode
if somebody puts it on their door the same way that they put
housenumbers on their door.