What is the steps for splitting childs table in postgresql? - sql

I have a Postgresql database, where many tables inherit from a common root table.
What are the SQL commands to use for removing the root table, but keeping data and final schema for each table?
Thanks.

A parent table cannot be dropped while any of its children remain. Neither can columns or check constraints of child tables be dropped or altered if they are inherited from any parent tables. If you wish to remove a table and all of its descendants, one easy way is to drop the parent table with the CASCADE option.
Source

Related

Creating mapping table in SQL

Can I create a mapping table out of a #temp table that I have created?
I need to create a mapping table that contains two different tables, but those tables do not have any field in common. However, I created a temp table with case statements and joining the table.
Now, I need to create a mapping table but I'm not sure if I can't create a mapping table out of a temp table.
What should I do?
You may want to look at creating a view instead. That will allow you to include case statements in it and you won't have to update your mapping table every time some data changes.
https://learn.microsoft.com/en-us/sql/t-sql/statements/create-view-transact-sql
If you include more information people will probably help further.

Parent & Child FK'ing to same table

What is best practise when a parent & child table both FK to the same table?
Parent > Child(ren)
CommonAttributes: Sex, Age, Height, Weight
Is it better to directly reference the common table:
CommonAttributes > Parent(s) > Child(ren)
&
CommonAttributes > Child(ren)
Or use a reference table:
RefTable: CommonAttributes_Id, Parent_Id(null), Child_Id(null)
I think the first method works OK (with regards to EF) but it is a bit of a circular reference. Is it better to use a reference table to define the constraints?
There are several approaches to this and the one you need depends on your business needs.
First, can a child record have more than one parent? For instance you might be modelling an organizational structure where an employee can have two supervisors. If this is true, then you have a one to many relationship and need a separate table for this model to work.
If you are guaranteed to have only one parent per child (but each parent might have a parent (building a hierarchy), then you can model this is one table. The table structure would include the Primary key, say UserID and then a nullable column for the parent such as ParentUserID. Then you can create the foreign key to the field in the same table.
ALTER TABLE dbo.Mytable ADD CONSTRAINT FK_Mytable _UserPArent FOREIGN KEY (ParentUserD) REFERENCESdbo.Mytable (UserID)
If you want to build a hierarchy in a query, you then use a recursive CTE to get it. See example here:
https://msdn.microsoft.com/en-us/library/ms186243.aspx
Another time you might want to build a separate table for the child parent relationship is if only a small portion of teh records in the main table would have parent child relationships. For instance suppose you had a people table that stored, sales reps and customers. Only sales reps would have a parent child relationship. So you would want a separate SalesRepHierarchy table to store it which woudl make querying more straightforward.
While in general you woudl want to create hierarchies in a recursive CTE, there are special cases when it might be faster to pre calculate the hierarchies. This is true if the hierarchy is frequently queried, the CTE performance is slow and you have control over how the hierarchy is built (preferably through an import of data only) and if it changes fairly rarely (you would not want to be rebuilding the hierarchy every minute, but a once a day import can be accommodated. This can greatly speed up and simply querying for the whole hierarchy, but is not recommended if the parent child relationships are created and changed constantly through the application.

PostgreSQL delete from parent table only

I have a table structure with partitioned tables, where a few child tables inherit from a common parent. How so I DELETE only from the parent table?
Long story short, I ended up with some data in the parent table, and this should've never happened but now I have to clean up the mess.
You can specify that only parent table matters, simply by using keyword 'ONLY':
DELETE FROM ONLY parent_table_name;
See: http://www.postgresql.org/docs/current/static/sql-delete.html

Inheritance with Hibernate

I need some advice on modelling my data tables. I need to apply inheritance hierarchy on my tables using SQL Server and Hibernate. Could anyone show me a basic example? It could be a tutorial on website too.
Cheers...
Set up the tables so that the derived table shares primary key with base table.
MS SQL Server isn't an object-orientated database, it is a relational database. It sounds like you should be using views over your base tables rather than duplicating columns.
Duplicating columns is unnecessary and would no doubt impact performance and maintenance would become a nightmare.
Maybe edit your question to include more details on what you are trying to achieve.
Another way is to duplicate attributes in child tables and make parent table as VIEW (that selects by common attributes from all children).
CREATE VIEW Parent
AS
SELECT ID, Name FROM Child1
UNION ALL
SELECT ID, Name FROM Child2 ...
The problem could be with ID that should be unique through all the child tables (using GUIDs is preferrable)

Inheritance in database?

Is there any way to use inheritance in database (Specifically in SQL Server 2005)?
Suppose I have few field like CreatedOn, CreatedBy which I want to add on all of my entities. I looking for an alternative way instead of adding these fields to every table.
There is no such thing as inheritance between tables in SQL Server 2005, and as noted by the others, you can get as far as getting help adding the necessary columns to the tables when you create them, but it won't be inheritance as you know it.
Think of it more like a template for your source code files.
As GateKiller mentions, you can create a table containing the shared data and reference it with a foreign key, but you'll either have to have audit hooks, triggers, or do the update manually.
Bottom line: Manual work.
PostgreSQL has this feature. Just add this to the end of your table definition:
INHERITS FROM (tablename[, othertable...])
The child table will have all the columns of its parent, and changes to the parent table will change the child. Also, everything in the child table will come up in queries to the parent table (by default). Unfortunately indices don't cross the parent/child border, which also means you can't make sure that certain columns are unique across both the parent and child.
As far as I know, it's not a feature used very often.
You could create a template in the template pane in Management Studio. And then use that template every time you want to create a new table.
Failing that, you could store the CreatedOn and CreatedBy fields in an Audit trail table referencing the original table and id.
Failing that, do it manually.
You could use a data modeling tool such as ER/Studio or ERWin. Both tools have domain columns where you can define a column template that you can apply to any table. When the domain changes so do the associated columns. ER/Studio also has trigger templates that you can build and apply to any table. This is how we update our LastUpdatedBy and LastUpdatedDate columns without having to build and maintain hundreds of trigger scripts.
If you do create an audit table you would have one row for every row in every table that uses the audit table. That could get messy. In my opinion, you're better off putting the audit columns in every table. You also may want to put a timestamp column in all of your tables. You never know when concurrency becomes a problem. Our DB audit columns that we put in every table are: CreatedDt, LastUpdatedBy, LastUpdatedDt and Timestamp.
Hope this helps.
We have a SProc that adds audit columns to a given table, and (optionally) creates a history table and associated triggers to track changes to a value. Unfortunately, company policy means I can't share, but it really isn't difficult to achieve.
If you are using GUIDs you could create a CreateHistory table with columns GUID, CreatedOn, CreatedBy. For populating the table you would still have to create a trigger for every table or handle it in the application logic.
You do NOT want to use inheritance to do this! When table B, C and D inherits from table A, that means that querying table A will give you records from B, C and D. Now consider...
DELETE FROM a;
Instead of inheritance, use LIKE instead...
CREATE TABLE blah (
blah_id serial PRIMARY KEY
, something text NOT NULL
, LIKE template_table INCLUDING DEFALUTS
);
Ramesh - I would implement this using supertype and subtype relationships in my E-R model. There are a few different physical options you have of implementing the relationships as well.
in O-R mapping, inheritance maps to a parent table where the parent and child tables use the same identifier
for example
create table Object (
Id int NOT NULL --primary key, auto-increment
Name varchar(32)
)
create table SubObject (
Id int NOT NULL --primary key and also foreign key to Object
Description varchar(32)
)
SubObject has a foreign-key relationship to Object. when you create a SubObject row, you must first create an Object row and use the Id in both rows