Reporting Services Cscading Parameter refresh

Question

I have googled it a lot and found that usually it cannot be done. I came across one of the hacks here:

http://www.bp-msbi.com/2011/04/ssrs-cascading-parameters-refresh-solved/

But its not working for me in ssrs 2005. Just wondering if anyone else tried it in 2005. Or is there any other hacks that can be tried.

As per this article the dependent parameter gets refreshed only when its values are invalidated by the selection in the first parameter. If we can invalidate the dependent parameter every time a parameter changes we will enforce a complete refresh. An easy way to do this is to attach a value such as a GUID obtained with the NEWID() T-SQL function.

So basically we want to introduce a fake parameter in between two real parameters. This fake parameter is supposed to return new values everytime as the storedproc behind it will add a guid to the resultset everytime that proc is called. So it forces complete refresh of the other parameters.

Now the main issue I am facing is :

Setting the default value of this fake parameter. For the available values the storedproc behind the fake param runs and it returns data in the format say : result1,result2_GUIDFROMSQL

Now it looks like the same storedproc is called again to set the defult value if i ask it to get the default value from query. But as the storedproc is run again new guid comes and so the old value cannot be found so its not being set as desired.

I just need to figure out a mechanism to pass this guid from introduced param to the next param.

Thats where I am failing.

My issue can simply be replicated by creating a param whose datasource is this querystring.

select getdate() id, @name nid 

So in this case how to set a default value for this param.

Answer 1

Below I'm going to present a detailed scenario and then I'll show the example implementation based on Visser and Abbi's answers.

Imagine that I have a report where each row is a project. A project has a Status column with values "In Progress" or "Complete" and a Project Manager column whose values are a person's name. Here are the rows in the table:

• Project A (Status = In Progress, Project Manager = Bob)
• Project B (Status = In Progress, Project Manager = Tom)
• Project C (Status = Complete, Project Manager = Jack)
• Project D (Status = Complete, Project Manager = Tom)
• Project E (Status = Complete, Project Manager = Jill)

I want to have 2 parameters on my report

1. Show Completed Projects? - This is a boolean parameter
   • When false will only show "In Progress" projects A & B
   • When true will show "In Progress" projects A & B in addition to "Complete" projects C, D, & E.
2. Project Manager - This is a multi-value text parameter whose options and default values will need to change based on the Show Completed Projects? parameter upon which it is dependent.
   • If Show Completed Projects? is set to false then only "Bob" and "Tom" options will show up because they are the project managers for the in progress projects Project A & B respectively.
   • If Show Completed Projects? is set to true then in addition to "Bob" and "Tom" you will also have "Jack" and "Jill" show up as options because they are project managers for the inactive projects Project C & Project E respectively.

Now for the implementation:

1. Show Completed Projects? parameter

   

2. Project Managers dataset query (See Visser and Abbi's answers for details on how this generates a key that will change based on the independent parameter and will force SSRS to reload the default values)

   SELECT 
       [ProjectManager_Key] = 
           pOuterAlias.[ProjectManager_Key] + '_' + 
           CAST(ROW_NUMBER() OVER(ORDER BY pOuterAlias.[ProjectManager_Key] DESC) AS NVARCHAR(MAX)),
       [ProjectManager] =  pOuterAlias.[ProjectManager]
   FROM 
       (
           SELECT 
               [ProjectManager_Key] = 
                   pInnerAlias.ProjectManager + '_' + 
                   CAST(ROW_NUMBER() OVER(ORDER BY pInnerAlias.ProjectManager ASC) AS NVARCHAR(MAX)),
               [ProjectManager] = pInnerAlias.ProjectManager
           FROM 
           (
               SELECT 
                   [ProjectManager] 
               FROM 
                   [dbo].[Project]
               WHERE
                   Status = 'In Progress' OR
                   @ShowCompletedProjects = 1
           ) pInnerAlias
       ) pOuterAlias
   ORDER BY 
       pOuterAlias.[ProjectManager]
   

3. Project Manager parameter

   • General

     

   • Available Values

     

   • Default Values

     

4. Projects dataset

   • Query

     SELECT 
         *
     FROM 
         [dbo].[Project]
     WHERE
     (
         Status = 'In Progress' OR
         @ShowCompletedProjects = 1
     ) AND
     Project Manager IN (@ProjectManager)
     

   • Parameters (Make sure to note the [@ProjectManager.Label] portion which will make it match the project on the actual project manager value from the database and not the key that we generated.

     

Answer 2

Finally I was able to resolve this. This link was a helpful start. http://www.optimusbi.com/2012/07/16/multilevel-cascading-select/

Basically what it does is : Writing parameter query in a manner so that the dependent parameter changes its value every time you change its parent parameter.

The query adds rownumber with a "_" preceeding it. So every time user selects other values the rownumber changes and hence the query resultset.

Then when using the resultset remove the text after the underscore to get the real codes.

Answer 3

There is a workaround to fix this issue for all situations.

Note that the previous answer provided, writing parameter the query in a manner so that the dependent parameter changes its value every time you change its parent parameter, works for some cases but not all cases. If the dependent parameter's "Available" values change as a result of another parameter AND the dependent parameter is visible, it works, If the dependent parameter is hidden or internal, or if the "Available" values do not change as a result of the other parameter, it will not work.

The foolproof workaround is to define a custom function in the report and call it in an expression for the dependent parameter. The custom function takes the parameter's (the parameter that the dependent parameter depends upon) value as an argument, and simply returns its value. Even though function is simply taking the value and returning it, SSRS does not inspect the code and assumes that the code could do anything (generate a random number, pull files from disk, etc ...). So SSRS calls the function every single time the value changes regardless of whether the dependent parameter's "Available" values change, and regardless of whether the dependent parameter is visible, hidden, or internal.

Something like this:

public function returnArg(ByVal TheArg As String) As String
  return TheArg
end function

Assume you have two parameters:

Parameter1 Parameter2

And that Parameter2 depends upon Parameter1.

Set Parameter2's value as an expression that includes the call to Parameter1 with the function, like:

=CODE.returnArg(Parameters!Parameter1.Value)

Now in this case Parameter2 simply displays the value in Parameter1, but this logic can be extended to more complex expressions and more than one parameter. As long as there is a CODE.returnArg(...) function call for each parameter in the expression, SSRS will always refresh the value.

Answer 4

@Abbi's second link is no longer valid. The article can now be found at: http://www.optimusinfo.com/multilevel-cascading-select/

Unfortunately the images are broken and the article is a little incomplete without them. There is one image I was able to locate in its moved location (I could not find the others even with trying appropriate variations): http://www.optimusinfo.com/wp-content/uploads/2012/07/Multilevel-Cascading-with-Select-All-18.jpeg

This gave me the last clue I needed to get the technique working: when you're setting up the dataset parameter properties, you need to manually specify the parameter value as e.g. [@City.Label] rather than the usual plain old [@City] (which corresponds to [@City.Value].

This is because the technique alters the "Value" to a custom value which can no longer be looked up in the DB! So you need to use the user-friendly, db-existing "Label" as the parameter. (Probably should be obvious but...) The rest is all pretty standard if you understand report builder.

Also note that this link as it stands doesn't work with multi value cascading parameters. However it can easily be modified to do so: just alter the parameter stored procs to join to a value splitting function. I adapted mine from the udf_splitvarible function given at SQL Server - In clause with a declared variable

The modification is fairly simple but I'll give an example. This is from the article (with terrible formatting fixed).

SELECT 
    l11.CompanyName1+'_'+ CAST(row_number() over( order by  l11.CompanyName1 desc) AS VARCHAR(50) )as CompanyName11
    ,l11.CompanyName
FROM 
(
    SELECT 
        l1.CompanyName+'_'+ CAST(row_number() over( order by  l1.CompanyName asc) AS VARCHAR(50) )as CompanyName1
        ,l1.CompanyName
    FROM 
    (
        SELECT DISTINCT CompanyName
        FROM Customers
        WHERE City IN(@City)
    )l1
)l11
ORDER BY l11.CompanyName ASC

Change:

SELECT DISTINCT CompanyName
FROM Customers
WHERE City IN(@City)

to:

SELECT DISTINCT CompanyName
FROM Customers
INNER JOIN udf_SplitVariable(@City,',') v ON City = v.Value

This all applies to SQL Server 2008/Report Builder 3 also.

I wasn't able to understand/apply @borkmark's answer. The limitations don't seem to apply to my case.