I need to update multiple rows in a Parts table when a field in another table changes and I want to use a trigger. The reason for the trigger is many existing application use and modify the data and I don't have access to all of them. I know some databases support a For Each Row in the trigger statement but I don't think Microsoft does.
Specificly I have two tables Parts and Categories.
Parts has Part#, Category_ID, Part_Name and Original and lots of other stuff
Category has Category_ID and Category_name.
Original is a concatenation of Category_Name and Part_Name separated by a ':'
For example Bracelets:BB129090
If someone changes the Category_Name (for excample from Bracelets to Bracelets), the Original field must be updated in every row of the Parts table. While this is an infrequent event it happens enough to cause trouble.
No Web and desktop applications uses Original
All Accounting application use only Original
It is my task to keep Accounting and the other application in sync.
I did not design the database and the company that wrote the accounting program will not change it.
Or another option: why don't you just create a view over those two tables, for your Accounting department, which contains this concatenated column:
CREATE VIEW dbo.AccountingView
AS
SELECT
p.PartNo, p.Part_Name, p.Category_ID,
c.Category_Name + ':' + p.PartName as 'Original'
FROM
Parts p
INNER JOIN
Category c ON p.Category_ID = c.Category_ID
Now your Accounting people can use this view for their reporting, it's always fresh, always up to date, and you don't have to worry about update and insert triggers and all those tricky things.....
Marc
The Original column violates 1NF, which is a very bad idea. You can either
Skip the column completely and concatenate it in each query (probably not the best solution, but I argue that it's probably better than the trigger).
Create a view over the table and have the Original column in the view (probably what I would do), or
Make Original a computed column, which is the best way if you want to create an index on it.
I guess in your case there is no need for a row-level trigger.
You can do something like
IF UPDATE(Category_Name)
UPDATE Parts
SET Original = inserted.Category_Name + ':' + Part_Name
FROM Parts
INNER JOIN inserted ON Parts.Category_ID = inserted.Category_ID
as an UPDATE trigger on the Category table.
If you really need per-row processing (say, of a stored procedure), you need a CURSOR or a WHILE loop over inserted.
If you can alter the table schemas, on option that you would have that would ensure that the Original column is always up to date, no matter what, is to make Original a computed column - a column that's computed from the Category_Name plus the Part_Name as needed.
For this, you need to create a stored function that will do that computation for you - something like this:
CREATE FUNCTION dbo.CreateOriginal(#Category_ID INT, #Part_Name VARCHAR(50))
RETURNS VARCHAR(50)
WITH SCHEMABINDING
AS BEGIN
DECLARE #Category_Name VARCHAR(50)
SELECT #Category_Name = Category_Name FROM dbo.Category
WHERE Category_ID = #Category_ID
RETURN #Category_Name + ': ' + #Part_Name
END
and then you need to add a column to your Parts table which will show the result of this function for each row in the table:
ALTER TABLE Parts
ADD Original AS dbo.CreateOriginal(Category_ID, Part_Name)
The main drawback is the fact that to display the column value, the function has to be called each time, for each row.
On the other hand, your data is always up to date and always guaranteed to be correct, no matter what. No triggers needed, either.
See if that works for you - depending on your needs and the amount of data you have, it might well perform just fine for you.
Marc
Related
I have to develop a DB stored procedure to be used in an autocomplete feature. If the user types "M" I should return people and groups with the name starting with "M".
I would like to do that in a way that the system should need to call only one stored procedure to avoid the cost of calling the DB twice.
My problem is that groups and people are in separate table (I just need their and and ID), this is what I would ideally do:
select ID, Name from people OR group where name like 'M%'
Of course this is not a valid SQL.
I thought about creating a temp table, making 2 selects (one for people and one groups), inserting in this table and returning it.
Is there a simpler/nicer way to implement it or is this the best approach?
Just perform a UNION ALL. Two separate queries, a temp table, multiple inserts, and then a 3rd query, doesn't that sound like a whole lot of unnecessary work?
CREATE PROCEDURE dbo.SearchTwoTables
#search_argument NVARCHAR(255) = N'M%'
AS
BEGIN
SET NOCOUNT ON;
SELECT ID, Name FROM dbo.People
WHERE Name LIKE #search_argument
UNION ALL
SELECT ID, Name FROM dbo.Group
WHERE Name LIKE #search_argument;
END
GO
Also I assume your Name column should support Unicode strings (and therefore be NVARCHAR), otherwise you may find an angry person when you mangle their name upon saving...
I need a two retrieve data from the same table but divided in different columns.
First table "PRODUCTS" has the following columns:
PROD_ID
PRO_TYPE_ID
PRO_COLOR_ID
PRO_WEIGHT_ID
PRO_PRICE_RANGE_ID
Second table "COUNTRY_TRANSLATIONS" has the following columns:
ATTRIBUTE_ID
ATT_LANGUAGE_ID
ATT_TEXT_ID
Third and last table "TEXT_TRANSLATIONS" has the following columns:
TRANS_TEXT_ID
TRA_TEXT
PRO_TYPE_ID, PRO_COLOR_ID, PRO_WEIGHT_ID and PRO_PRICE_RANGE_ID are all integers and are found back in the column ATTRIBUTE_ID multiple times (depending on howmany translations are available). Then ATT_TEXT_ID is joined with TRANS_TEXT_ID from the TEXT_TRANSLATIONS table.
Basically I need to run a query so I can retreive information from TEXT_TRANSLATIONS multiple times. Right now I get an error saying that the correlation is not unique.
The data is available in more then 20 languages, therefore the need to work with intergers for each of the attributes.
Any suggestion on how I should build up the query? Thank you.
Hopefully, you're on an RDBMS that supports CTEs (pretty much everything except mySQL), or you'll have to modify this to refer to the joined tables each time...
WITH Translations (attribute_id, text)
as (SELECT c.attribute_id, t.tra_text
FROM Country_Translations c
JOIN Text_Translations t
ON t.trans_text_id = c.att_text_id
WHERE c.att_language_id = #languageId)
SELECT Products.prod_id,
Type.text,
Color.text,
Weight.text,
Price_Range.text
FROM Products
JOIN Translations as Type
ON Type.attribute_id = Products.pro_type_id
JOIN Translations as Color
ON Color.attribute_id = Products.pro_color_id
JOIN Translations as Weight
ON Weight.attribute_id = Products.pro_weight_id
JOIN Translations as Price_Range
ON Price_Range.attribute_id = Products.pro_price_range_id
Of course, personally I think the design of the localization table was botched in two ways -
Everything is in the same table (especially without an 'attribute type' column).
The language attribute is in the wrong table.
For 1), this is mostly going to be a problem because you now have to maintain system-wide uniqueness of all attribute values. I can pretty much guarantee that, at some point, you're going to run into 'duplicates'. Also, unless you've designed your ranges with a lot of free space, the data values are non-consecutive for type; if you're not careful there is the potential for update statements being run over the wrong values, simply because the start and end of the given range belong to the same attribute, but not every value in the range.
For 2), this is because a text can't be completely divorced from it's language (and country 'locale'). From what I understand, there are parts of some text that are valid as written in multiple languages, but mean completely different things when read.
You'd likely be better off storing your localizations in something similar to this (only one table shown here, the rest are an exercise for the reader):
Color
=========
color_id -- autoincrement
cyan -- smallint
yellow -- smallint
magenta -- smallint
key -- smallint
-- assuming CYMK palette, add other required attributes
Color_Localization
===================
color_localization_id -- autoincrement, but optional:
-- the tuple (color_id, locale_id) should be unique
color_id -- fk reference to Color.color_id
locale_id -- fk reference to locale table.
-- Technically this is also country dependent,
-- but you can start off with just language
color_name -- localized text
This should make it so that all attributes have their own set of ids, and tie the localized text to what it was localized to directly.
I have a Users table that contains dozens of columns like date of birth, year of vehicle owned, make and model of the vehicle, color and many other personal fields unrelated to the vehicle
There's also a 2nd table called Coupons that needs to be designed in a way to support a qualification like "user qualifies if younger than 30 yrs old", "user qualifies if vehicle is greater than 10 yrs old", "user qualifies if vehicle color is green".
When a user logs in, I need to present all coupons the user qualifies for. The problem that I'm having is that the coupon qualifications could be numerous, could have qualifiers like equal, greater than or less than and may have different combinations.
My only solution at this point is to store the actual sql string within one of the coupons table columns like
select * from Users where UserId = SOME_PLACEHOLDER and VehicleYear < 10
Then I could execute the sql for each coupon row and return true or false. Seems very inefficient as I would potentially have to execute 1000s of sql statements for each coupon code.
Any insight, help is appreciated. I do have server-side code where I could potentially be able to do looping.
Thank you.
Very difficult problem. Seems like users will be added at high volume speed, with coupons at a fairly regular frequency.
Adding SQL to a table to be used dynamically is workable - at least you'll get a fresh execution plan - BUT your plan cache may balloon up.
I have a feeling that running a single coupon for all users is probably likely to be your highest performing query because it's one single set of criteria which will be fairly selective on users first and total number of coupons is small, whereas running all coupons for a single user is separate criteria for each coupon for that user. Running all coupons for all users may still perform well, even though it's effectively a cross join first - I guess it is just going to depend.
Anyway, the case for all coupons for all users (or sliced either way, really) will be something like this:
SELECT user.id, coupon.id
FROM user
INNER JOIN coupon
ON (
CASE WHEN <coupon.criteria> THEN <coupon.id> -- code generated from the coupon rules table
CASE WHEN <coupon.criteria> THEN <coupon.id> -- etc.
ELSE NULL
) = coupon.id
To generate the coupon rules, you can relatively easily do the string concatenation in a single swipe (and you can combine an individual rule lines design for a coupon with AND with a further inner template):
DECLARE #outer_template AS varchar(max) = 'SELECT user.id, coupon.id
FROM user
INNER JOIN coupon
ON (
{template}
ELSE NULL
) = coupon.id
';
DECLARE #template AS varchar(max) = 'CASE WHEN {coupon.rule} THEN {coupon.id}{crlf}';
DECLARE #coupon AS TABLE (id INT, [rule] varchar(max));
INSERT INTO #coupon VALUES
(1, 'user.Age BETWEEN 20 AND 29')
,(2, 'user.Color = ''Yellow''');
DECLARE #sql AS varchar(MAX) = REPLACE(
#outer_template
,'{template}',
REPLACE((
SELECT REPLACE(REPLACE(
#template
,'{coupon.rule}', coupon.[rule])
, '{coupon.id}', coupon.id)
FROM #coupon AS coupon
FOR XML PATH('')
), '{crlf}', CHAR(13) + CHAR(10)));
PRINT #sql;
// EXEC (#sql);
There's ways to pretty that up - play with it here: https://data.stackexchange.com/stackoverflow/q/115098/
I would consider adding computed columns (possibly persisted and indexed) to assist. For instance, age - non-persisted computed column will likely perform better than a scalar function.
I would consider batching this with a table which says whether a coupon is valid for a user and when it was last validated.
Seems like ages can change and a user can become valid or invalid for a coupon as their birthday passes.
When a user logs in you could spawn a background job to update their coupons. On subsequent logons, there won't be any need to update (since it's not likely to change until the next day or a triggering event).
Just a few ideas.
I would also add that you should have a way to test a coupon before it is approved to ensure there are no syntax errors (since the SQL is ad hoc or arbitrary) - this can be done relatively easily - perhaps a test user table (test_user as user in the generated code template instead) is required to contain pass and fail rows and the coupon rule points to those. Not only does the EXEC have to work - the rows it returns should be the expected and only the expected rows for that coupon.
This is not an easy problem. Here are some quick ideas that may help depending on your domain requirements:
Restrict the type of criteria you will be filtering on so that you can use dynamic or non-dynamic sql to execute them efficiently. For example if you are going to only have integers between a range of min and max values as a criteria then the problem becomes simpler. (You only need to know the field name, and the min max values to describe a criterian, not the full where statement.)
Create a number of views which expose the attributes in a helpful way. Then perform queries against those views -- or have those views pre-select in some way. For example, an age group view that has a field which can contain the values < 21, 21-30, 30-45, >45. Then your select just needs to return the rows from this view that match these strings.
Create a table which stores the results of running your criteria matching query (This can be run off line by a back ground process). Then for a given user check for membership by looking where in the table this user's ID exists.
Thinking about this some more I realize all my suggestions are based on one idea.
A query for an individual user will work faster overall if you first perform an SQL query against all users and cache that result in some way. If every user is reproducing queries against the whole dataset you will lose efficiency. You need some way to cache results and reuse them.
Hope this helps -- comment if these ideas are not clear.
My first thought on an approach (similar to Hogan's) would be to test for coupon applicability at the time the coupon is created. Store those results in a table (User_Coupons for example). If any user data is changed, your system would then retest any changed users for which coupons are applicable to them. At coupon creation (or change) time it would only check versus that coupon. At use creation (or change) time it would only check versus that user.
The coupon criteria should be from a known set of possible criteria and any time that you want to add a new type of criteria, it would possibly involve a code change. For example, let's say that you have a table set up similar to this:
CREATE TABLE Coupon_Criteria (
coupon_id INT NOT NULL,
age_minimum SMALLINT NULL,
age_maximum SMALLINT NULL,
vehicle_color VARCHAR(20) NULL,
...
CONSTRAINT PK_Coupon_Criteria PRIMARY KEY CLUSTERED (coupon_id)
)
If you wanted to add the ability to base a coupon on vehicle age then you would have to add a column to the table and likewise you would have to adjust your search code. You would use NULL values to indicate that the criteria is unused for that coupon.
An example query for the above table:
SELECT
CC.coupon_id
FROM
Users U
INNER JOIN Coupon_Criteria CC ON
(CC.age_maximum IS NULL OR dbo.f_GetAge(U.birthday) <= age_maximum) AND
(CC.age_minimum IS NULL OR dbo.f_GetAge(U.birthday) >= age_minimum) AND
(CC.vehicle_color IS NULL OR U.vehicle_color = CC.vehicle_color) AND
...
This can get unwieldy if the number of possible criteria gets to be very large.
Another possibility would be to save the coupon criteria in XML and have a business object for your application use that to determine eligibility. It could use the XML to generate a proper query against the User table (and any other necessary tables).
Here's another possibility. Each criteria could be given a query template which you could append to your queries. This would just involve updates to the data instead of DDL and could have good performance. It would involve dynamic SQL.
CREATE TABLE Coupons (
coupon_id INT NOT NULL,
description VARCHAR(2000) NOT NULL,
...
CONSTRAINT PK_Coupons PRIMARY KEY CLUSTERED (coupon_id)
)
CREATE TABLE Coupon_Criteria (
coupon_id INT NOT NULL,
criteria_num SMALLINT NOT NULL,
description VARCHAR(50) NOT NULL,
code_template VARCHAR(500) NOT NULL,
CONSTRAINT PK_Coupon_Criteria PRIMARY KEY CLUSTERED (coupon_id, criteria_num),
CONSTRAINT FK_Coupon_Criteria_Coupon FOREIGN KEY (coupon_id) REFERENCES Coupons (coupon_id)
)
INSERT INTO Coupons (coupon_id, description)
VALUES (1, 'Young people save $200 on yellow vehicles!')
INSERT INTO Coupon_Criteria (coupon_id, criteria_num, description, code_template)
VALUES (1, 1, 'Young people', 'dbo.Get_Age(U.birthday) <= 20')
INSERT INTO Coupon_Criteria (coupon_id, criteria_num, description, code_template)
VALUES (1, 2, 'Yellow Vehicles', U.vehicle_color = ''Yellow''')
You could then build a query by simply concatenating all of the criteria for any given coupon. The big downside to this one is that it's only one-directional. Given a coupon you can easily find who is qualified for it, but given a user you cannot find all coupons for which they are eligible except by going through all of the coupons. My guess is that the second is what you'd probably be most interested in unfortunately. Maybe this will give you some other ideas though.
For example, you could potentially have it work the other way by having a set number of criteria in a table and for the coupon/criteria linking table indicate whether or not that criteria is active. When querying you could then include that in your query. In other words, the query would look something like:
WHERE
(CC.is_active = 0 OR <code from the code column>) AND
The querying gets very complex though since you either need to join once for every possible criteria or you need to query to compare the number of active requirements for a coupon versus the number that are fulfilled. That is possible in SQL, but it's similar to working with an EAV model - which is basically what this turns into: a variation on an EAV model (yuck)
In our application user can create different lists (like sharepoint) for example a user can create a list of cars (name, model, brand) and a list of students (name, dob, address, nationality), e.t.c.
Our application should be able to query on different columns of the list so we can't just serialize each row and save it in one row.
Should I create a new table at runtime for each newly created list? If this was the best solution then probably Microsoft SharePoint would have done it as well I suppose?
Should I use the following schema
Lists (Id, Name)
ListColumns (Id, ListId, Name)
ListRows (Id, ListId)
ListData(RowId, ColumnId, Value)
Though a single row will create as many rows in list data table as there are columns in the list, this just doesn't feel right.
Have you dealt with this situation? How did you handle it in database?
what you did is called EAV (Entity-Attribute-Value Model).
For a list with 3 columns and 1000 entries:
1 record in Lists
3 records in ListColumns
and 3000 Entries in ListData
This is fine. I'm not a fan of creating tables on-the-fly because it could mess up your database and you would have to "generate" your SQL queries dynamically. I would get a strange feeling when users could CREATE/DROP/ALTER Tables in my database!
Another nice feature of the EAV model is that you could merge two lists easily without droping and altering a table.
Edit:
I think you need another table called ListRows that tells you which ListData records belong together in a row!
Well I've experienced something like this before - I don't want to share the actual table schema so lets do some thought exercises using some of the suggested table structures:
Lets have a lists table containing a list of all my lists
Lets also have a columns table containing the metadata (column names)
Now we need a values table which contains the column values
We also need a rows table which contains a list of all the rows, otherwise it gets very difficult to work out how many rows there actually are
To keep things simple lets just make everything a string (VARCAHR) and have a go at coming up with some queries:
Counting all the rows in a table
SELECT COUNT(*) FROM [rows]
JOIN [lists]
ON [rows].list_id = [Lists].id
WHERE [Lists].name = 'Cars'
Hmm, not too bad, compared to:
SELECT * FROM [Cars]
Inserting a row into a table
BEGIN TRANSACTION
DECLARE #row_id INT
DECLARE #list_id INT
SELECT #list_id = id FROM [lists] WHERE name = 'Cars'
INSERT INTO [rows] (list_id) VALUES (#list_id)
SELECT #row_id = ##IDENTITY
DECLARE #column_id INT
-- === Need one of these for each column ===
SELECT #column_id = id FROM [columns]
WHERE name = 'Make'
AND list_id = #list_id
INSERT INTO [values] (column_id, row_id, value)
VALUES (#column_id, #row_id, 'Rover')
-- === Need one of these for each column ===
SELECT #column_id = id FROM [columns]
WHERE name = 'Model'
AND list_id = #list_id
INSERT INTO [values] (column_id, row_id, value)
VALUES (#column_id, #row_id, 'Metro')
COMMIT TRANSACTION
Um, starting to get a little bit hairy compared to:
INSERT INTO [Cars] ([Make], [Model}) VALUES ('Rover', 'Metro')
Simple queries
I'm now getting bored of constructing tediously complex SQL statements so maybe you can have a go at coming up with equivalent queries for the followng statements:
SELECT [Model] FROM [Cars] WHRE [Make] = 'Rover'
SELECT [Cars].[Make], [Cars].[Model], [Owners].[Name] FROM [Cars]
JOIN [Owners] ON [Owners].id = [Cars].owner_id
WHERE [Owners].Age > 50
SELECT [Cars].[Make], [Cars].[Model], [Owners].[Name] FROM [Cars]
JOIN [Owners] ON [Owners].id = [Cars].owner_id
JOIN [Addresses] ON [Addresses].id = [Owners].address_id
WHERE [Addresses].City = 'London'
I hope you are beginning to get the idea...
In short - I've experienced this before and I can assure you that creating a database inside a database in this way is definitely a Bad Thing.
If you need to do anything but the most basic querying on these lists (and literally I mean "Can I have all the items in this list please?"), you should try and find an alternative.
As long as each user pretty much has their own database I'll definitely recommend the CREATE TABLE approach. Even if they don't I'd still recommend that you at least consider it.
Perhaps a potential solution would be the creating of lists can involve CREATE TABLE statements for those entities/lists?
It sounds like the db structure or schema can change at runtime, or at the user's command, so perhaps something like this might help?
User wants to create a new list of an entity never seen before. Call it Computer.
User defines the attributes (screensize, CpuSpeed, AmountRAM, NumberOfCores)
System allows user to create in the UI
system generally lets them all be strings, unless can tell when all supplied values are indeed dates or numbers.
build the CREATE scripts, execute them against the DB.
insert the data that the user defined into that new table.
Properly coded, we're working with the requirements given: let users create new entities. There was no mention of scale here. Of course, this requires all input to be sanitized, queries parameterized, actions logged, etc.
The negative comment below doesn't actually give any good reasons, but creates a bit of FUD. I'd be interested in addressing any concerns with this potential solution. We haven't heard about scale, security, performance, or usage (internal LAN vs. internet).
You should absolutely not dynamically create tables when your users create lists. That isn't how databases are meant to work.
Your schema is correct, and the pluralization is, in my opinion, also correct, though I would remove the camel case and call them lists, list_columns, list_rows and list_data.
I would further improve upon your schema by skipping rows and columns tables, they serve no purpose. Simply have a row/column number attached to each cell, and keep things sparse: Don't bother holding empty cells in the database. You retain the ability to query/sort based on row/column, your queries will be (potentially very much) faster because the number of list_cells will be reduced, and you won't have to do any crazy joining to link your data back to its table.
Here is the complete schema:
create table lists (
id int primary key,
name varchar(25) not null
);
create table list_cells (
id int primary key,
list_id int not null references lists(id)
on delete cascade on update cascade,
row int not null,
col int not null,
data varchar(25) not null
);
It sounds like you might have Sharepoint already deployed in your environment.
Consider integrating your application with Sharepoint, and have it be your datastore. No need to recreate all the things you like about Sharepoint, when you could leverage it.
It'd take a bit of configuring, but you could call SP web services to CRUD your list data for you.
inserting list data into Sharepoint via web services
reading SP lists via web services
Sharepoint 2010 can also expose lists via OData, which would be simple to consume from any application.
I've set up a view which combines all the data across several tables. Is there a way to write this so that only columns which contain non-null data are displayed, and those columns which contain all NULL values are not included?
ADDED:
Sorry, still studying and working on my first big project so every day seems to be a new experience at the minute. I haven't been very clear, and that's partly because I'm not sure I'm going about things the right way! The client is an academic library, and the database records details of specific collections. The view I mentioned is to display all the data held about an item, so it is bringing together tables on publication, copy, author, publisher, language and so on. A small number of items in the collection are papers, so have additional details over and above the standard bibliographic details. What I didn't want was a user to get all the empty fields relating to papers if what was returned only consisted of books, therefore the paper table fields were all null. So I thought perhaps there would be a way to not show these. Someone has commented that this is the job of the client application rather than the database itself, so I can leave this until I get to that phase of the project.
There is no way to do this in sql.
CREATE VIEW dbo.YourView
AS
SELECT (list of fields)
FROM dbo.Table1 t1
INNER JOIN dbo.Table2 t2 ON t1.ID = t2.FK_ID
WHERE t1.SomeColumn IS NOT NULL
AND t2.SomeOtherColumn IS NOT NULL
In your view definition, you can include WHERE conditions which can exclude rows that have certain columns that are NULL.
Update: you cannot really filter out columns - you define the list of columns that are part of your view in your view definition, and this list is fixed and cannot be dynamically changed......
What you might be able to do is us a ISNULL(column, '') construct to replace those NULLs with an empty string. Or then you need to handle excluding those columns in your display front end - not in the SQL view definition...
The only thing I see you could do is make sure to select only those columns from the view that you know aren't NULL:
SELECT (list of non-null fields) FROM dbo.YourView
WHERE (column1 IS NOT NULL)
and so forth - but there's no simple or magic way to select all columns that aren't NULL in one SELECT statement...
You cannot do this in a view, but you can do it fairly easily using dynamic SQL in a stored procedure.
Of course, having a schema which shifts is not necessarily good for clients who consume the data, but it can be efficient if you have very sparse data AND the consuming client understands the varying schema.
If you have to have a view, you can put a "header" row in your view which you can inspect client-side on the first row in your loop to see if you want to not bother with the column in your grid or whatever, you can do something like this:
SELECT * FROM (
-- This is the view code
SELECT 'data' as typ
,int_col
,varchar_col
FROM TABLE
UNION ALL
SELECT 'hdr' as typ
-- note that different types have to be handled differently
,CASE WHEN COUNT(int_col) = 0 THEN NULL ELSE 0 END
,CASE WHEN COUNT(varchar_col) = 0 THEN NULL ELSE '' END
FROM TABLE
) AS X
-- have to get header row first
ORDER BY typ DESC -- add other sort criteria here
If we're reading your question right, there won't be a way to do this in SQL. The output of a view must be a relation - in (over-)simplified terms, it must be rectangular. That is, each row must have the same number of columns.
If you can tell us more about your data and give us some idea of what you want to do with the output, we can perhaps offer more positive suggestions.
In general, add a WHERE clause to your query, e.g.
WHERE a IS NOT NULL AND b IS NOT NULL AND c IS NOT NULL
Here, a b c are your column names.
If you are joining tables together on potentially NULL columns, then use an INNER JOIN, and NULL values will not be included.
EDIT: I may have misunderstood - the above filters out rows, but you may be asking to filter out columns, e.g. you have several columns and you only want to display columns that contain at least one null value across all the rows you are returning. Using dynamic SQL offers a solution, since the set columns varies depending upon your data.
Here's a SQL query that builds another SQL query containing the appropriate columns. You could run this query, and then submit it's result as another query. It assumes 'pk' is some column that is always non-null, e.g. a primary key - this means we can prefix additional row names with a comma.
SELECT CONCAT("SELECT pk"
CASE (count(columnA)) WHEN 0 THEN '' ELSE ',columnA' END,
CASE (count(columnB)) WHEN 0 THEN '' ELSE ',columnB' END,
// etc..
' FROM (YourQuery) base')
FROM
(YourQuery) As base
The query works using Count(column) - the aggregate function ignores NULL values, and so returns 0 for a column consisting entirely of NULLs. The query builder assumes that YourQuery uses aliases to ensure there no duplicate column names.
While you cant put this into a view, you could wrap it up as a stored procedure that copies the data to another table - the result table. You may also set up a trigger so that the result table is updated whenever the base tables change.
I suspect what's going on is that an end user is running CrystalReports and complaining about all the empty columns that have to be removed manually.
It would actually be possible to create a stored procedure that would create a view on the fly, leaving out dataless columns. But then you would have to run this proc before using the view.
Is that acceptable?