In SQL Server 2000/2005,
Is it possible to force the default value to be written to already existing rows when adding a new column to a table without using NOT NULL on the new column?
You need two statements. First create the column with not null. Then change the not null constraint to nullable
alter table mytable add mycolumn varchar(10) not null default ('a value')
alter table mytable alter column mycolumn varchar(10) null
I understand your question, but you are saying that for future records, NULL (unknown, indeterminate or whatever your semantics are) is acceptable (but if it is left off in an insert, there will be a default), but that for all the existing data, you are going to go ahead and assign it the default.
I would have to look hard at this situation and ask why you are even going to allow NULLs in future records at all - given none of the historical records will have it, and there is a default in place for future records.
I doubt it.
http://msdn.microsoft.com/en-us/library/ms190273(SQL.90).aspx
The approach recommended by Microsoft is as follows (taken from the url above)
UPDATE MyTable SET NullCol = N'some_value' WHERE NullCol IS NULL
ALTER TABLE MyTable ALTER COLUMN NullCOl NVARCHAR(20) NOT NULL
ALTER TABLE {TABLENAME}
ADD {COLUMNNAME} {TYPE} {NULL|NOT NULL}
CONSTRAINT {CONSTRAINT_NAME} DEFAULT {DEFAULT_VALUE}
[**WITH VALUES]**
WITH VALUES can be used to store the default value in the new column for each existing row in the table.
more detail on MSDN link .
https://msdn.microsoft.com/en-in/library/ms190273.aspx
Related
Currently I have a column named CreatedDate in a Shipper table I created. It is nullable. My task is to change the created date to be a required field, not allow nulls, and have a default of GetDate(). This has to be done in a single query... Keep in mind there is no data in my table yet. I've tried the following code and I can't seem to get it to work. This is a homework assignment and I'm only looking for guidance. Any help would be greatly appreciated. Thanks in advance.
USE Business
ALTER TABLE Shipper
ALTER COLUMN CreatedDate date NOT NULL
CONSTRAINT CNDefaultCreatedDate
DEFAULT GETDATE() For CreatedDate;
Two different statements, I assume this is SQL Server:
--Add the default
ALTER TABLE [Shipper] --What table
ADD CONSTRAINT [def_createddate] --Give the constraint a name
DEFAULT GETDATE() FOR [CreatedDate]; --default of what for which column
--Set to not allow NULL
ALTER TABLE [Shipper] --What table
ALTER COLUMN [CreatedDate] DATE NOT NULL; --altering what column and either NULL or NOT NULL
Understand adding a default will not update existing data. I know you mentioned your table does not have data, but in the future the null values must be updated to some value before the ALTER COLUMN NOT NULL is allowed.
Here's reference to the MS documentation ALTER TABLE
If the column did not already exist you can add it, set it not null and then even update existing rows in one statement:
--If Shipper did not already have the column CreatedDate
ALTER TABLE [Shipper]
ADD [CreatedDate] DATE NOT NULL
CONSTRAINT [def_createddate] DEFAULT GETDATE()
WITH VALUE --if column is nullable use this and it will update existing records with the default. It column is NOT NULL this is applied by default.
Have you tried this:
USE Business
ALTER TABLE Shipper
ALTER COLUMN CreatedDate date NOT NULL DEFAULT GETDATE();
Is NULL or NOT NULL necessary for Adding a column in SQL Server? (If you are going to run UPDATE statement after)
I tested without it locally, and it seems to work fine both in SQL and on the website; but I wanted to make sure before running release/production.
I looked up some other articles, including microsoft website. Some show it with, some without. SO articles say some benefits of NULL, like if you have information that may be added later. But assuming I am going to run UPDATE to add values after, will it matter?
I am guessing it does not matter from what I've tested and read.
ALTER TABLE [dbo].[MyTable] ADD NewColumn varchar(150);
UPDATE [dbo].[MyTable] SET NewColumn ='Math' WHERE ID = 1
UPDATE [dbo].[MyTable] SET NewColumn ='Science' WHERE ID = 2
You can skip it, and by default the column will be created as NULL. However it is more legible if you indicate it explicitly.
Keep in mind that if your table already has data, you CANNOT add the column as NOT NULL. For this, you should firstly add the column as NULL, then UPDATE the values with non-null valid data and then alter column to NOT NULL.
Edit: Assuming the default behavior of the sql server when adding columns.
You don't have to write it, when you skip it column will be created as 'NULL'.
Following statements are equal:
ALTER TABLE [dbo].[MyTable] ADD NewColumn varchar(150);
ALTER TABLE [dbo].[MyTable] ADD NewColumn varchar(150) NULL;
If you want to add a column and DB engine should guarantee you that this new column will be always populated you can add a NOT NULL column with a default value.
Other option will be to change the column to NOT NULL after the update.
I'd make sure you have something declarative. And if you are setting as "not null," be sure you have a default value or use can suffer LOTS of data loss if data previously exists. It makes me nervous because this an ALTER vs a CREATE.
I want to have a column in which the default timestamp is added when an insertion is done. I have done that in MySQL using choosing default timestamp option. How do I do that in SQL Server?
ALTER TABLE table_name ADD InsertTime DATETIME DEFAULT GETDATE()
ALTER TABLE [table] ADD [CreationTimeStampUTC] DATETIME DEFAULT(GETUTCDATE())
Or you could do the same with GETDATE(), which I think I've seen more in real-world code. Note, of course, that GETDATE() will be valued with the current server time of the SQL Server, and may or may not match up with the current time of the web server if you have one, or particularly the client's machine. Obviously there's no issue with that, just make sure you're aware of the implications ahead of time. You could also store this timestamp as a DATETIMEOFFSET to rid yourself of some of those concerns, but since you're speaking explicitly of on-server default values, I think UTC is perfectly sufficient.
What you're looking for is a default constraint. How you add it depends on whether the column already exists or not. If it doesn't, you can add the column and the constraint in one shot like so:
alter table dbo.yourTable
add [NewColumn] datetime
constraint [DF_NewColumn] default (getdate());
If the column already exists, you can attach a default to it like so:
alter table dbo.yourTable
add constraint [DF_ExistingColumn] default (getdate()) for [ExistingColumn];
Finally, if the table doesn't yet exist, you can add the constraint to the table definition:
create table dbo.yourTable (
NewColumn datetime not null constraint [DF_NewColumn] default (getdate())
)
Note: in all cases, I gave the constraint a name. You can choose not to, but then you'll eventually get a huffy DBA asking why her database comparison is a mess because the constraint names don't match up between dev and prod. :)
I have a table [Product] with a column [CreateTime] datetime null, and is has some data already.
How can I set the column [CreateTime] 's default value to getdate(), and make the new added data to have a default value getdate() for column [CreateTime].
You cannot change a default - you will need to first drop it, and then recreate it.
In order to drop it, you need to know its name, and then use
ALTER TABLE dbo.Product
DROP CONSTRAINT yourOldDefaultConstraint
Once you've done that, you can add a new default constraint, and in order to apply it to existing rows, use the "WITH VALUES" part:
ALTER TABLE dbo.Product
ADD CONSTRAINT NewDefaultConstraintName
DEFAULT GetDate() FOR CreateTime WITH VALUES
Oops - sorry, the "WITH VALUES" only seems to work if you create a DEFAULT constraint at the time you create the table, or if you add the column - it doesn't seem to get applied to an existing column.
In this case you would just have to follow your ALTER TABLE statement with something like this:
UPDATE dbo.T_Product
SET CreateTime = GETDATE()
WHERE CreateTime IS NULL
That should do the trick, too!
Marc
I have a table that has several nullable integer columns. This is undesirable for several reasons, so I am looking to update all nulls to 0 and then set these columns to NOT NULL. Aside from changing nulls to 0, data must be preserved.
I am looking for the specific SQL syntax to alter a column (call it ColumnA) to "not null". Assume the data has been updated to not contain nulls.
Using SQL server 2000.
First, make all current NULL values disappear:
UPDATE [Table] SET [Column]=0 WHERE [Column] IS NULL
Then, update the table definition to disallow "NULLs":
ALTER TABLE [Table] ALTER COLUMN [Column] INTEGER NOT NULL
I had the same problem, but the field used to default to null, and now I want to default it to 0. That required adding one more line after mdb's solution:
ALTER TABLE [Table] ADD CONSTRAINT [Constraint] DEFAULT 0 FOR [Column];
You will have to do it in two steps:
Update the table so that there are no nulls in the column.
UPDATE MyTable SET MyNullableColumn = 0
WHERE MyNullableColumn IS NULL
Alter the table to change the property of the column
ALTER TABLE MyTable
ALTER COLUMN MyNullableColumn MyNullableColumnDatatype NOT NULL
For Oracle 11g, I was able to change the column attribute as follows:
ALTER TABLE tablename MODIFY columnname datatype NOT NULL;
Otherwise abatichev's answer seemed good. You can't repeat the alter - it complains (at least in SQL Developer) that the column is already not null.
this worked for me:
ALTER TABLE [Table]
Alter COLUMN [Column] VARCHAR(50) not null;
As long as the column is not a unique identifier
UPDATE table set columnName = 0 where columnName is null
Then
Alter the table and set the field to non null and specify a default value of 0
In case of FOREIGN KEY CONSTRAINT... there will be a problem if '0' is not present in the column of Primary key table. The solution for that is...
STEP1:
Disable all the constraints using this code :
EXEC sp_msforeachtable "ALTER TABLE ? NOCHECK CONSTRAINT all"
STEP2:
RUN UPDATE COMMAND (as mentioned in above comments)
RUN ALTER COMMAND (as mentioned in above comments)
STEP3:
Enable all the constraints using this code :
exec sp_msforeachtable #command1="print '?'", #command2="ALTER TABLE ? WITH CHECK CHECK CONSTRAINT all"
this seems simpler, but only works on Oracle:
ALTER TABLE [Table]
ALTER [Column] NUMBER DEFAULT 0 NOT NULL;
in addition, with this, you can also add columns, not just alter it.
It updates to the default value (0) in this example, if the value was null.
In my case I had difficulties with the posted answers. I ended up using the following:
ALTER TABLE table_name CHANGE COLUMN column_name column_name VARCHAR(200) NOT NULL DEFAULT '';
Change VARCHAR(200) to your datatype, and optionally change the default value.
If you don't have a default value you're going to have a problem making this change, as default would be null creating a conflict.
Making column not null and adding default can also be done in the SSMS GUI.
As others have already stated, you can't set "not null" until all
the existing data is "not null" like so:
UPDATE myTable SET myColumn = 0
Once that's done, with the table in design view (right click on
table and click "design view"), you can just uncheck the Allow
Nulls columns like so:
Still in design view with the column selected, you can see the
Column Properties in the window below and set the default to 0 in there as well like so:
Let's take an example:
TABLE NAME=EMPLOYEE
And I want to change the column EMPLOYEE_NAME to NOT NULL. This query can be used for the task:
ALTER TABLE EMPLOYEE MODIFY EMPLOYEE.EMPLOYEE_NAME datatype NOT NULL;
For the inbuilt javaDB included in the JDK (Oracle's supported distribution of the Apache Derby) the below worked for me
alter table [table name] alter column [column name] not null;
You can change the definition of existing DB column using following sql.
ALTER TABLE mytable modify mycolumn datatype NOT NULL;
First make sure the column that your changing to not does not have null values
select count(*) from table where column's_name is null
Impute the missing values. you can replace the nulls with empty string or 0 or an average or median value or an interpolated value. It depends on your back fill strategy or forward fill strategy.
Decide if the column values need to be unique or non-unique. if they need to be unique than add an unique constraint. Otherwise, see if performance is adequate or if you need to add an index.