After hunting around on various forums for almost an hour, I've come to the conclusion that SQL server is slightly stupid about simple arithmetic.
I am attempting to utilize a function which, until recently seemed to work just fine. Upon changing out some of the values for a different set of information on the form in use, I get the odd behavior ahead.
The problem is that it is giving me the incorrect result as based on an excel spreadsheet formula.
The formula looks like this:
=IF(D8=0,0,(((D8*C12-C16)*(100-C13)/100+C16)/D8)+(C18*D8))
My SQL looks like this:
(((#DaysBilled * #ContractRate - #ActualPlanDed) * (100 - #InsCover) / 100 + #ActualPlanDed) / #DaysBilled) + (#CoPay * #DaysBilled)
Filling the variables with the given data looks like this:
(((11 * 433 - 15) * (100 - 344) / 100 + 15) / 11) + (15 * 11)
Even stranger, if I use the numbers above (adding .00 to the end of each value) manually in the server environment, it gives me -11405.1200000000
With the values I am giving, it should come out 166.36. Unfortunately, I am getting -886.83
Here is the entire function and how it is called:
ALTER FUNCTION Liability
(
#ClientGUID CHAR(32),
#RecordGUID CHAR(32),
#Type CHAR(3)
)
RETURNS DECIMAL(18,2) AS
BEGIN
DECLARE #ReturnValue decimal(18,2);
DECLARE #DaysBilled int;
DECLARE #ContractRate decimal(18,2);
DECLARE #ActualPlanDed decimal(18,2);
DECLARE #InsCover decimal(18,2);
DECLARE #CoPay decimal(18,2);
IF (#Type = 'RTC')
BEGIN
SELECT #DaysBilled = RTCDaysBilled,
#ContractRate = CAST(REPLACE(REPLACE(ContractRateRTC, ' ',''),'$', '') AS DECIMAL(6,2)),
#ActualPlanDed = RTCActualPlanDed,
#InsCover = InsRTCCover,
#CoPay = RTCCoPay
FROM AccountReconciliation1
WHERE #ClientGUID = tr_42b478f615484162b2391ef0b2c35ddc
AND #RecordGUID = tr_abb4effa0d9c4fe98c78cb4d2e21ba5d
END
IF (#Type = 'PHP')
BEGIN
SELECT #DaysBilled = PHPDaysBilled,
#ContractRate = CAST(REPLACE(REPLACE(ContractRatePHP, ' ',''),'$', '') AS DECIMAL(6,2)),
#ActualPlanDed = PHPActualPlanDed,
#InsCover = InsPHPCover,
#CoPay = PHPCoPay
FROM AccountReconciliation1
WHERE #ClientGUID = tr_42b478f615484162b2391ef0b2c35ddc
AND #RecordGUID = tr_abb4effa0d9c4fe98c78cb4d2e21ba5d
END
IF (#Type = 'IOP')
BEGIN
SELECT #DaysBilled = IOPDaysBilled,
#ContractRate = CAST(REPLACE(REPLACE(ContractRateIOP, ' ',''),'$', '') AS DECIMAL(6,2)),
#ActualPlanDed = IOPActualPlanDed,
#InsCover = InsIOPCover,
#CoPay = IOPCoPay
FROM AccountReconciliation1
WHERE #ClientGUID = tr_42b478f615484162b2391ef0b2c35ddc
AND #RecordGUID = tr_abb4effa0d9c4fe98c78cb4d2e21ba5d
END
IF (#DaysBilled <> 0)
BEGIN
SET #ReturnValue = (((#DaysBilled * #ContractRate - #ActualPlanDed)
*
(100 - #InsCover) / 100 + #ActualPlanDed)
/
#DaysBilled
)
+
(#CoPay * #DaysBilled)
END
ELSE
BEGIN
SET #ReturnValue = 0;
END
RETURN #ReturnValue;
END
It is called by running a select statement from our front-end, but the result is the same as calling the function from within management studio:
SELECT dbo.Liability('ClientID','RecordID','PHP') AS Liability
I have been reading about how a unary minus tends to break SQL's math handling, but I'm not entirely sure how to counteract it.
One last stupid trick with this function: It must remain a function. I cannot convert it into a stored procedure because it must be used with our front-end, which cannot utilize stored procedures.
Does SQL server even care about the parentheses? Or is it just ignoring them?
The calculation is correct, it differes of course if you are using float values
instead of integers.
For (((11 * 433 - 15) * (100 - 344) / 100 + 15) / 11) + (15 * 11)
a value around -886.xx depending in which places integers/floats are used is correct,
What makes you believe it should be 166.36?
Related
Situation: I have a column where each cell can have up to 5 delimiters. However, it's possible that there are none.
Objective: How do i handle errors such as :
Invalid length parameter passed to the LEFT or SUBSTRING function.
in the case that it cannot find the specified delimiter.
Query:
declare #text VARCHAR(111) = 'abc-def-geeee-ifjf-zzz'
declare #start1 as int
declare #start2 as int
declare #start3 as int
declare #start4 as int
declare #start_index_reverse as int
set #start1 = CHARINDEX('-',#text,1)
set #start2 = CHARINDEX('-',#text,charindex('-',#text,1)+1)
set #start3 = CHARINDEX('-',#text,charindex('-',#text,CHARINDEX('-',#text,1)+1)+1)
set #start4 = CHARINDEX('-',#text,charindex('-',#text,CHARINDEX('-',#text,CHARINDEX('-',#text,1)+1)+1)+1)
set #start_index_reverse = CHARINDEX('-',REVERSE(#text),1)
select
LEFT(#text,#start1-1) AS Frst,
SUBSTRING(#text,#start1+1,#start2-#start1-1) AS Scnd,
SUBSTRING(#text,#start2+1,#start3-#start2-1) AS Third,
SUBSTRING(#text,#start3+1,#start4-#start3-1)AS Third,
RIGHT(#text,#start_index_reverse-1) AS Lst
In this case my variable includes 5 delimiters and so my query works but if i removed one '-' it would break.
XML support in SQL Server brings about some unintentional but useful tricks. Converting this string to XML allows for some parsing that is far less messy than native string handling, which is very far from awesome.
DECLARE #test varchar(111) = 'abc-def-ghi-jkl-mnop'; -- try also with 'abc-def'
;WITH n(x) AS
(
SELECT CONVERT(xml, '<x>' + REPLACE(#test, '-', '</x><x>') + '</x>')
)
SELECT
Frst = x.value('/x[1]','varchar(111)'),
Scnd = x.value('/x[2]','varchar(111)'),
Thrd = x.value('/x[3]','varchar(111)'),
Frth = x.value('/x[4]','varchar(111)'),
Ffth = x.value('/x[5]','varchar(111)')
FROM n;
For a table it's almost identical:
DECLARE #foo TABLE ( col varchar(111) );
INSERT #foo(col) VALUES('abc-def-ghi-jkl-mnop'),('abc'),('def-ghi');
;WITH n(x) AS
(
SELECT CONVERT(xml, '<x>' + REPLACE(col, '-', '</x><x>') + '</x>')
FROM #foo
)
SELECT
Frst = x.value('/x[1]','varchar(111)'),
Scnd = x.value('/x[2]','varchar(111)'),
Thrd = x.value('/x[3]','varchar(111)'),
Frth = x.value('/x[4]','varchar(111)'),
Ffth = x.value('/x[5]','varchar(111)')
FROM n;
Results (sorry about the massive size, seems this doesn't handle 144dpi well):
add a test before your last select
then you should decide how to handle the other case (when one of start is 0)
You can also refer to this link about splitting a string in sql server
which is uses a loop and can handle any number of delimiters
if #start1>0 and #start2>0 and #start3>0 and #start4>0
select LEFT(#text,#start1-1) AS Frst,
SUBSTRING(#text,#start1+1,#start2-#start1-1) AS Scnd,
SUBSTRING(#text,#start2+1,#start3-#start2-1) AS Third,
SUBSTRING(#text,#start3+1,#start4-#start3-1)AS Third,
RIGHT(#text,#start_index_reverse-1) AS Lst
I have a stored procedure:
ALTER PROCEDURE [dbo].[spUpdateOrInsertNotification]
#ContentJsonHash BINARY(32)
AS
DECLARE #NotificationId INT;
SET #NotificationId = (SELECT #NotificationId
FROM dbo.tblNotifications n
WHERE n.ContentJsonHash = #ContentJsonHash);
IF #NotificationId IS NOT NULL
BEGIN
-- Increment Count
END
ELSE
BEGIN
-- Insert new row.
END
It's supposed to check if the Hash already exists and if it does, increment the count for the row, otherwise insert the row. However, it never finds the Hash and the corresponding NotificationId. NotificationId is always null.
If I run it twice, passing it the same data (a C# array byte[32]). It never finds the same NotificationId and I end up with duplicate entries being put in.
e.g.
NotificationId | ContentJsonHash
9 0xB966C33517993003D789EDF78DA20C4C491617F8F42F76F48E572ACF8EDFAC2A
10 0xB966C33517993003D789EDF78DA20C4C491617F8F42F76F48E572ACF8EDFAC2A
Can I not do comparisons on Binary(n) fields like this WHERE n.ContentJsonHash = #ContentJsonhash ?
The C# code:
using (var conn = new SqlConnection(Sql.ConnectionString))
{
await conn.OpenAsync();
using (var cmd = new SqlCommand(Sql.SqlUpdateOrInsertNotification, conn))
{
cmd.CommandType = CommandType.StoredProcedure;
cmd.Parameters.AddWithValue("#Source", notificationMessage.Source);
cmd.Parameters.AddWithValue("#Sender", notificationMessage.Sender);
cmd.Parameters.AddWithValue("#NotificationType", notificationMessage.NotificationType);
cmd.Parameters.AddWithValue("#ReceivedTimestamp", notificationMessage.Timestamp);
cmd.Parameters.AddWithValue("#ContentJSon", notificationMessage.NotificationContent);
cmd.Parameters.AddWithValue("#ContentJsonHash", notificationMessage.ContentHashBytes);
await cmd.ExecuteNonQueryAsync();
}
}
I've also tried calling the stored procedure from SQL like this:
exec dbo.spUpdateOrInsertNotification 'foo', 'bar', 0,
'2017-12-05 15:23:41.207', '{}',
0xB966C33517993003D789EDF78DA20C4C491617F8F42F76F48E572ACF8EDFAC2A
Calling this twice returns 2 rows :(
I can do this, which works, hard coding the binary field I want to check
select *
from dbo.tblNotifications
where ContentJsonhash = 0xB966C33517993003D789EDF78DA20C4C491617F8F42F76F48E572ACF8EDFAC2A
Binary comparisons can be tricky. If you are using a true binary column, I believe length also comes into play. So even if those bytes are the same, and the lengths differ, the comparison would be false. An easy way is to convert these to strings:
alter procedure [dbo].[spUpdateOrInsertNotification]
#ContentJsonHash BINARY(32)
AS
DECLARE #NotificationId INT;
SET #NotificationId = (SELECT NotificationId
FROM dbo.tblNotifications n
WHERE convert(varchar(32), n.ContentJsonHash, 2) = convert(varchar(32), #ContentJsonHash, 2));
IF #NotificationId IS NOT NULL
BEGIN
-- Increment Count
END
ELSE
BEGIN
-- Insert new row.
END
I had an # where I shouldn't have had an ampersand.
SET #NotificationId = (SELECT #NotificationId
FROM dbo.tblNotifications n
WHERE convert(varchar(32), n.ContentJsonHash, 2) = convert(varchar(32), #ContentJsonHash, 2));
Should be
SET #NotificationId = (SELECT NotificationId
FROM dbo.tblNotifications n
WHERE convert(varchar(32), n.ContentJsonHash, 2) = convert(varchar(32), #ContentJsonHash, 2));
I feel so stupid for not noticing this sooner :(
I found this code snipped (Source):
CREATE PROCEDURE rec_fib(n INT, OUT out_fib INT)
BEGIN
DECLARE n_1 INT;
DECLARE n_2 INT;
IF (n=0) THEN
SET out_fib=0;
ELSEIF (n=1) then
SET out_fib=1;
ELSE
CALL rec_fib(n-1,n_1);
CALL rec_fib(n-2,n_2);
SET out_fib=(n_1 + n_2);
END IF;
END
This code works with MySQL. In how far do I have to modify it to run on DB2? I cannot seem to find a running minimal example of an recursive stored procedure for DB2.
This works for me: (I haven't done more than make it work, so alternative coding could also work.)
First, add these two lines:
DECLARE n_3 INT;
DECLARE n_4 INT;
Then modify this small section:
ELSE
set n_3 = n - 1;
set n_4 = n - 2;
CALL rec_fib(n_3,n_1);
CALL rec_fib(n_4,n_2);
That's all. Runs on IBM i 6.1 DB2 UDB.
The following code is from SQL tips for DB2, written by Serge Rielau
CREATE OR REPLACE FUNCTION Fib(n INTEGER) RETURNS DECIMAL(31, 0)
BEGIN
DECLARE res DECIMAL(31, 0);
CASE WHEN n = 0 THEN
SET res = 0;
WHEN n = 1 THEN
SET res = 1;
WHEN n > 1 THEN
BEGIN
DECLARE stmt STATEMENT;
PREPARE stmt FROM 'SET ? = Fib(? - 1) + Fib(? - 2)';
EXECUTE stmt INTO res USING n, n;
END;
ELSE
SIGNAL SQLSTATE '78000' SET MESSAGE_TEXT = 'Bad input';
END CASE;
RETURN res;
END;
/
For more information, please check the source page of this code: https://www.ibm.com/developerworks/community/blogs/SQLTips4DB2LUW/entry/recursive_sql_pl?lang=en
This is my stored procedure:
CREATE PROCEDURE _spCalc
(#Num01 decimal(18,0), #Num02 decimal(18,0), #Num03 decimal(18,0))
AS
BEGIN
DECLARE #Var01 float, DECLARE #Var02 float, DECLARE #Var03 float
SET #Var01 = #Num01 * 1000
SET #Var02 = #Num02 - ((POWER(6.53 * #Var01, 0.5)) / (POWER(#Num03, 0.5)))
SET #Var03 = (1 - 3.4680733 * LOG(#Var02) + 1.8779192 * POWER(LOG(#Var02), 2))
INSERT INTO _myTable(Num01, Num02, Num03, Num04)
VALUES (#Num01, #Num02, #Num03, ((#Var02 + #Var03 + #Num01) * 1000))
END
My question is, how can I test and know the result of each variable that I declared? I want to compare the result with the calculation on excel format. Just want to make sure, the result is the same.
Cheers,
we can use "print #var01". repeat for different variable names before or after calculation.this shows up in messages section of results.
or
we can use " select #var01 as var01,#var02 as var02,#var03 as var03" before your insert statement.this gives you a table like output during execution
you can extend the print or select with other variables as needed.
I could not add further comments for some reasons.
"try the calculations as separate "select" before assigning to variable"
select #Num01 * 1000 as #Var01
select #Num02 - ((POWER(6.53 * #Var01, 0.5)) / (POWER(#Num03, 0.5))) as #Var02
select (1 - 3.4680733 * LOG(#Var02) + 1.8779192 * POWER(LOG(#Var02), 2))
substitute the values that u pass to the procedure accordingly
I'm having another issue with a SQL Function that I had written. If I run it not in a function and use declared variables, then it works perfectly. But as soon as I put it into a function and run it, nothing appears, its empty.
I cannot put this into a stored procedure, it needs to be in a function.
the code is
select * from [MYTABLE]
where MajorGroupID = #MajorGroupID
and ((#Status = 0 and (
Inactive = 0)
))
or MajorGroupID = #MajorGroupID and (#Status = 1 and (Inactive = 0 or Inactive = 1))
I am not really familiar with functions, I can do basic things with functions but when it comes to adding logic to it. If I was allowed to use stored procedures then I wouldn't be having problems.
This is MSSQL and using SQL Server 2010.
EDIT, Added complete function
CREATE FUNCTION [dbo].[WT_FN_GET_MYTABLE_By_MajorGroupID_Inactive]
(
#MajorGroupID varchar,
#Status int
)
RETURNS TABLE
AS
RETURN
select * from [MYTABLE]
where MajorGroupID = #MajorGroupID
and ((#Status = 0 and (
Inactive = 0)
))
or MajorGroupID = #MajorGroupID and (#Status = 1 and (Inactive = 0 or Inactive = 1))
You don't provide a length for #MajorGroupID varchar, so it is going to be 1 by default, at which point it will not find anything in the table.
Provide a length, e.g. #MajorGroupID varchar(30).