Column break in nested matrix

Question

Hi guys,
I created a report of some quality test results containing table with nested matrix for details. I need the
sub-matrix  to be grouped by row and by
column and limit the number of result columns in row (to fit the width of report neatly). I went trough the articles about doing column group break already:
http://social.msdn.microsoft.com/Forums/sqlserver/en-US/32f28407-e1ca-457e-92fd-d292e32dde4e/limit-no-of-columns-in-ssrs-matrix-report?forum=sqlreportingservices 
http://social.msdn.microsoft.com/Forums/sqlserver/en-US/ea9d795b-8d17-41d2-a1d7-a4069ebb4539/forum-faq-how-do-i-achieve-column-break-in-a-matrix?forum=sqlreportingservices (I have used the 2nd workaround)
My solution without the column count limit feature works perfectly. But when trying to break, the report seems to associate the column groups related to another parent table row, because the number of result grouped columns in row is
less or equal to the set number (=4) as follows:
Test type Test
A A1
Parameter A1-1 A1-2 A1-3 A1-4 // OK - break made after 4 columns of A1
Min x x x x
Max y y y y
Measured z z z z
Parameter A1-5 A1-6 // OK - break made after test type A1 (?Bug cause: "column count 2 for now")
Min x x
Max y y
Measured z z
A2
Parameter A2-1 // OK - break made after test type A2 (?Bug cause: "column count 3 for now")
Min x
Max y
Measured z
B B1
Parameter B1-1 // Not OK - break already made after 1st column (?Bug cause: "column count 4 for now:"=>BREAK)
Min x
Max y
Measured z
Parameter B1-2 B1-3 B1-4 // OK - break made after test type B1
Min x x x
Max y y y
Measured z z z
C C1
Parameter C1-1 C1-2 C1-3 C1-4 // OK - break made after 4 columns of C1
Min x x x x
Max y y y y
Measured z z z z
Parameter C1-5 C1-6 // OK - break made after test type C1
Min x x
Max y y
Measured z z
I have modified the proposed function script for my purposes - set the reset of ID counter in the header of  parent row test, set the column ID  as concatenation of all level values (test type, test and watched parameter),  played with ordering
of data and extra grouping, but with no success.
Does anybody have a suggestion what is going on and how to solve this issue?
Thanks,
Marion
PS: I'm sorry for the table illustration (done the best that I could). As a newbie I'm not allowed to post with picture. 
Edit: Do you need some extra information to help? Does somebody have an idea how to achieve the same design with column break some other way?

Answer 1

Hi,
The column break needs to reset when it starts a new row grouping.  The customer code therefore needs parameters for the row grouping and for the column grouping.
The code below is slightly different to a lot of examples.
Creating hashtables would keep a record of every column value, so for large reports, this would use a small but significant amount of memory.  It only needs to know when it changes.  I've also used static variables as the scope only needs to be
the function and not global, plus this makes it a single module of code which helps on larger projects.
The ceiling calculation has been replaced with some integer maths to prevent odd rounding issues that cause hard to trace/reproduce bugs.  E.G. a 4 column should be 4-4-4-4-4 but 16 /4 should be 4 but might be
4.000000000000001.  Ceilling then returns 5 so you end up with a column jumping to the next line :- 4-3-5-4-4.
Note that everything that used to calculate the row grouping should be concatenated as a single parameter for the most robust solution.  In the example the sub group is unique for the entire dataset so is not an issue, but if a sub group is only unique
in the group, there is a small change of a miscalculation.
Here's my solution:
In the code
Function GroupFunction(
ByVal NewGroup as string ,
ByVal NewColumn as string ,
ByVal ColumnBreak as Integer ) As Integer
static OldGroup as string = ""
static OldColumn as string = ""
static RecordCount as integer = 0
if (OldGroup <> NewGroup ) then
OldGroup = NewGroup
OldColumn = ""
RecordCount = RecordCount + ( ColumnBreak - RecordCount Mod ColumnBreak )
end if
if (OldColumn <> NewColumn) then
OldColumn = NewColumn
RecordCount = RecordCount +1
end if
GroupFunction =RecordCount +
( ColumnBreak - (RecordCount - 1) Mod ColumnBreak )
End Function
In the report
= Code.GroupFunction( Fields!MainGroup.Value & ":" & Fields!SubGroup.Value   , Fields!Parameter.Value , 4 )
If you want a different number of columns, you can change the 4 to a report parameter.
Output
A
A1
Value Type
A1-01
A1-02
A1-03
A1-04
Max
2
5
8
11
Measured
3
6
9
12
Min
1
4
7
10
Value Type
A1-05
A1-06
A1-07
A1-08
Max
14
1
1
1
Measured
15
1
1
1
Min
13
16
1
1
Value Type
A1-09
A1-10
Max
1
1
Measured
1
1
Min
1
1
A
A2
Value Type
A2-01
Max
1
Measured
1
Min
1
B
B1
Value Type
B1-01
B1-02
B1-03
B1-04
Max
1
1
1
1
Measured
1
1
1
1
Min
1
1
1
1
Value Type
B1-05
Max
1
Measured
1
Min
1
C
C1
Value Type
C1-01
C1-02
C1-03
C1-04
Max
1
1
1
1
Measured
1
1
1
1
Min
1
1
1
1
Value Type
C1-05
C1-06
Max
1
1
Measured
1
1
Min
1
1
Best Regards
Fergus