数据库表设计问题？

Question

I am creating a table with name as a column such that there are multiple rows with same name and corresponding to every name there are other values (columns).

I have to design the table such that there are two set of values default and latest . So user may enquire to get either default/latest set and may edit table to make particular set of values as default replacing the previous defult set.

I am thinking of implementing this by keeping a flag column where: 0 means this row is default set and 1 means the latest set and 2 means just a general row.

TABLE:

NAME            ADDRESS         CITY         FLAG

Hansen          Timoteivn 10    Sandnes      1   (LATEST)   
Hansen          street 13       texas        0   (DEFAULT)
Svendson        Borgvn 23       Sandnes      1   (LATEST)
Svendson        street 14       colaba       2   (GENERAL)
Svendson        street 15       Houston      0   (DEFAULT)
Pettersen       Storgt 20       Stavanger    0   (DEFAULT)
Pettersen       Storgt 21       texas        1   (LATEST)

I am thinking of implementing it like this:

alter table TABLE_NAME add flag number CONSTRAINT <constraintname>
   CHECK (flag IN (0, 1, 2));

Is there a better approach to do this apart from keeping a flag ?? Am i doing it wrong??

Answer 1

Is there a better approach to do this apart from keeping a flag ??

Typically a default setting will be show by a flag column but not information about the latest.

To store latest information a table will include 2 additional timestamp columns: Created and Modified (and you might have two additional columns showing who eg CreatedBy, ModifiedBy).

If you have these auditing columns then figuring out the latest is just a matter of looking at these timestamps depending on if latest means most recent created or most recent modified.

Answer 2

Is there a better approach to do this apart from keeping a flag ?? Am i doing it wrong??

If you can't distinguish "default" data from "latest" data any other way, you have to either add a column or add a table.

But I don't understand why you'd choose to use an integer the way you're using one. You're representing two distinct values with three distinct integers.

Answer 3

With your proposal there is nothing from stopping you from storing multiple defaults or latest. You also do not have a proper primary key.

A better (more relational) solution would be to a) add some kind of version to your table and b) create another table with the PK (name + which version is default copy) with those two columns as a primary & foreign key. "Latest" would simply be the row with the highest version, not a flag. This would give you better data integrity. However it will allow the situation where a given key may not have a default.

To enforce that you have to redesign your table to have a parent record (name), some child records(other stuff) and a FK on your parent that points to the default. Depending on whether your DB supports deferred constraints (I don't know MySQL) this may not be possible due to the circular nature.

Answer 4

Your schema is missing a table for storing unique names.

If you create one, then you can have an address table that references it, and given that there can be only one "latest" and one "default" row per "name" you can add columns to the name table to link to the latest and default rows of address.

Of course if latest means "last added to the table" then adding a timestamp for the creation of the addresses means that you can infer the latest row from that.