Steve,
The format you presented is pretty much what most GIS/Cadastral systems use. Although it seems pretty straight forward, it can get very complicated. I had a project working with a utility company in Arizona, and half of the street names/addresses were in Spanish where the format is backwards (or is our system backwards?). I had to write a program to parse the addresses to separate the different components for cadastral compliance; it is a good thing that my Spanish is still good and even then, many columns were left empty as it was not feasible to parse the address and have the same result when put together. I am in favor of simplicity. I normally have one column Address1 where the 5 components you mentioned are stored, and one column Address2 where any additional information, such as Apt. No. go. This works for 99.99% of addresses in this country. The duplication, blank columns and additional storage space do not bother me as storage is dirt cheap. Javier. Javier Valencia, PE O: 913-829-0888 H: 913-397-9605 C: 913-915-3137 From: [email protected] [mailto:[email protected]] On Behalf Of Wills, Steve Sent: Thursday, February 23, 2012 9:50 AM To: RBASE-L Mailing List Subject: [RBASE-L] - RE: Too relational? William, may I encourage you to take a gander at that USPS document, http://pe.usps.gov/cpim/ftp/pubs/pub28/pub28.pdf, beginning at Section 231, as it breaks-down the shape of the address, specifically the Address, into their discrete parts : Ø Primary Address (aka: Line 1) o Primary Address Number o Predirectional o Street Name o Suffix o Postdirectional o Example: 123 South Main Street West (You can see that simple concatenation creates output for all of us users, but you might also begin to see why a standardized record layout could be useful in some applications: lets say I wanted to search for Main AND Street, excluding any circle, avenue, cove, parkway, etc.) Ø Secondary address (aka: Line 2, all that Suite, Apartment, Unit stuff.) I have seen such a record-structure several times, often related to government/public records, such as property identification/location (as in titles and 911 or who owns this address and how do I get a fire truck to it), so I think its some sort of standard. I cant say when this structure was created or last revised, but, regardless of what anyone might think about the USPS, theyve been in the address business for over 200 years. And, since I had to double-check the PDF to be sure I wasnt telling you a lie, I discovered that its actually CHOCKED FULL of seemingly-esoteric-but-potentially-useful information about any and everything to do with any address to which the USPS delivers. IOW, nothing about Canada, Germany, Ghana, New Zealand, etc, but I that other link I sent yesterday might cover that topic to the same excruciating degree of detail! Another $0.02, Steve From: [email protected] [mailto:[email protected]] On Behalf Of William Stacy Sent: Thursday, February 23, 2012 9:22 AM To: RBASE-L Mailing List Subject: [RBASE-L] - RE: Too relational? I think people can not only have several postal addresses, but they can have several telephone #s, multiple work lines, faxes, etc. even multiple e-mails are popular. But you're right about the shape thing. I'm thinking that the street name and the city,state zip line probably belong in their own tables as lookups. I think Folsom CA 95630 should only exist once in my entire database, not the thousands of times it now does. But that brings me back to my title question of this thread... On Thu, Feb 23, 2012 at 4:38 AM, Bill Downall <[email protected]> wrote: William, This is a design dilemma. Addresses are definitely a separate table, because a person can have multiple addresses, and because the "shape" of the data is different, (street address, city, state, postal code, country). But do the phones and emails and twitter accounts link back to the people, or to the addresses? I tend to go with linking to the people, with phone types broken down in "home phone" "mobile phone" "work phone", etc. Bill On Thu, Feb 23, 2012 at 7:25 AM, William Stacy <[email protected]> wrote: Now this is interesting. Do you include Postal contact types the same way, in the same table? Are these 2 columns part of a personal demographic table, or a separate table. If the latter, how do you link them up with the personal table? TIA On Wed, Feb 22, 2012 at 5:59 PM, Bill Downall <[email protected]> wrote: I cannot see the future as well as you, Mike. But my more recent designs do not have any columns with the letters p-h-o-n-e in a column name. There is a column for ContactType, and another for ContactValue. I could someday add a new contact type of ipv6, in addition to existing types of email, mobile, work, google voice, twitterID, etc. No schema change needed. Bill On Feb 22, 2012 5:46 PM, "Mike Byerley" <[email protected]> wrote: I started using nnn.nnn.nnnn for phone numbers anticipating at some time sub ipv6, phones will just be IP numbers. Just a guess though. ----- Original Message ----- From: "Bill Downall" <[email protected]> To: "RBASE-L Mailing List" <[email protected]> Sent: Wednesday, February 22, 2012 11:26 AM Subject: [RBASE-L] - RE: Too relational? It's nice to see Professor Wills here! You know a topic like this would get him going. Bill, in my mind, a basic reason to normalize fully is to create a database that is least likely to need either schema changes or awkward exception-handling down the road. If you do not normalize, and you provide room for 3 phone numbers, some day you will have to put the fourth phone number in the comments, or change the schema to allow for 4 phone numbers. Schema changes are expensive, because all forms and reports and procedures and eeps and views and rules and triggers and applications that relate to that data may have to be changed, too, and cannot be done by users through "settings", but have to be done by programmers. Putting the data in the "wrong" place like the comments means people won't find that data with a normal search or query. There are other good reasons to normalize, like not "wasting" columns that are usually blank, and not having to search three or five columns instead of one (For example, to determine what customer might have sent us an incomplete or garbled fax message or credit card transaction where all we know is that their address is "345 Main Street"). But avoiding future expensive schema changes is the main one. Bill On Wed, Feb 22, 2012 at 11:02 AM, Wills, Steve <[email protected]> wrote: > Too relational is a state that is rarely achieved, IMHO. I think your > issue/question often and I like the direction of your thinking. I guess > that thinking about such makes me a little twisted to some. I also own > my own barcode-scanner - well enough about my predilections!**** > > -- William Stacy, O.D. Please visit my website by clicking on : http://www.folsomeye.net -- William Stacy, O.D. Please visit my website by clicking on : http://www.folsomeye.net
