asked on

Bit mask error codes

Hi All

We have an application that returns error code data as bit mapped.

ie
1 - error 1
2 - error 2
4 - error 3
8 - error 4
16 error 5

etc

If the app returne either a 1,2,4,8... I kow the error, but more oft than not it is a combination of errors ie

9 - error 1 and error 4.

Does anyone have an algorithm to convert these decimals back ito the error codes. I have a list of codes.

Ideally I'd like to do this on SQL Server, but code behind would be ok.

Andy

ToddBeaulieu

You need BITWISE operators

declare @ERROR_1 int
declare @ERROR_2 int
declare @ERROR_3 int
declare @ERROR_4 int
 
set @ERROR_1 = 1
set @ERROR_2 = 2
set @ERROR_3 = 4
set @ERROR_4 = 8
 
declare @err int
set @err = @ERROR_1 | @ERROR_3
print '@err = ' + convert(varchar, @err)
 
if (@err & @ERROR_1 = @ERROR_1)
  print 'ERROR 1'
if (@err & @ERROR_2= @ERROR_2)
  print 'ERROR 2'
if (@err & @ERROR_3 = @ERROR_3)
  print 'ERROR 3'
if (@err & @ERROR_4 = @ERROR_4)
  print 'ERROR 4'

Open in new window

Aneesh

You need to use a logical AND operation

DECLARE @Error int
SELECT @Error = 9
SELECT CASE WHEN @Error&1 =0 THEN 'False' ELSE 'True' END as 'Error 1',
CASE WHEN @Error&2 =0 THEN 'False' ELSE 'True' END as 'Error 2',
CASE WHEN @Error&4 =0 THEN 'False' ELSE 'True' END as 'Error 3',
CASE WHEN @Error&8 =0 THEN 'False' ELSE 'True' END as 'Error 4'

Andy Green

ASKER

Not sure I understand your solution, or maybe it's my descrition of the problem.
I get a decimal value from the external app, which when converted to binary, equates to an error code, where ever there ia a 1.
The posible values go up to 1024. ie
1,2,4,8,16,32,64,128,256,512,1024 11 possible errors.
I cant test for every posible combination, how do I take the decimal, convert to binary and map to the error codes.
Andy

ToddBeaulieu

Why wouldn't you test for every condition? You have to, in order to map to the error codes, right?
You do a bit-wise AND against a test code and if the result = the test code, then that error code is contained in the map.

Andy Green

ASKER

I see what you are saying, yes I agree I have to test aginst each of the 11, but I dont see how I get from the decimal to the actual bits, that represent the errors.
I'll take a look over the weekend. Friday at 5 club now (well I'm late) and that means beer.
Andy

ASKER CERTIFIED SOLUTION

ToddBeaulieu

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

Andy Green

ASKER

Thanks - got it now. It was the conversion into binary that stumped me and it does it for you implicitly.

Andy

Anthony Perkins

>>I dont see how I get from the decimal to the actual bits<<
You may not have any control over this, but there is no need to use decimals for this, an integer will suffice. In fact if all you have are 11 variations then a smallint will do.

ToddBeaulieu

"decimal" did not refer to a data type, but rather to "base 10".

Anthony Perkins

Ah, good point. I should read more carefully.