Re: Enforcing uniqueness on [real estate/postal] addresses

From: Peter Devoy <peter(at)3xe(dot)co(dot)uk>
To: "Peter J(dot) Holzer" <hjp-pgsql(at)hjp(dot)at>
Cc: pgsql-general(at)lists(dot)postgresql(dot)org
Subject: Re: Enforcing uniqueness on [real estate/postal] addresses
Date: 2020-05-12 20:55:56
Message-ID: CABoFc_jWOMgKxm7yRTLRO_ZPyNi_L0LhxCvbM2CPGKFt56b88Q@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

>Is is possible to have two entries which have the same
>address_identifier_general, street and postcode, but different
>descriptions?

Unfortunately, yes. The data comes from gov't systems to
regulate the development/alteration of arbitrary pieces of property and
those pieces do not always have a postal address. E.g. a farmer may
one year apply to erect a wind turbine in "field north of Foo Cottage"
and the next year apply to demolish "barnhouse west of Foo Cottage".

Now, I know what you are thinking, there is a normalization opportunity and you
may well be right. However, the problem does exist in some of the other fields
too and I am already facing a fair amount of join complexity in my schema so I
am trying to figure out my options :)

>(What is an address_identifier_general, btw?)
Address identifier composed by numbers and/or characters. I'm using the
terminology from the EU's "INSPIRE Data Specification on Addresses" Guidelines.

I haven't yet had the opportunity to try out the above suggestions but I will
post again when I have.

In response to

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Karsten Hilbert 2020-05-12 21:31:36 Re: Enforcing uniqueness on [real estate/postal] addresses
Previous Message Adrian Klaver 2020-05-12 19:30:17 Re: ESQL/C: a ROLLBACK rolls back a COMMITED transaction