UNIX sort

[lokesh_dhiran]

hi,
While using the sort stage i have to come to know that the unix sort option always sorts Character wise. That is between 10 and 7 10 comes before 7 in ascending order.

Please tell me if this patch has been rectified or is there any other problem.

[novice_pgr]

unix sort is by default character wise . to sort it on numeric value ..it has to be sort -n

hi,
While using the sort stage i have to come to know that the unix sort option always sorts Character wise. That is between 10 and 7 10 comes before 7 in ascending order.

Please tell me if this patch has been rectified or is there any other problem.

[kumar_s]

It depends on how you define the column data type.
If you assign the column as Integer you will get as expected.
If you assign as varchar or character, you will get the result from datastage as you instructed the datastage to do.

[lokesh_dhiran]

I will differ with you on this. My key is integer but stil character sort is taking place

It depends on how you define the column data type.
If you assign the column as Integer you will get as expected.
If you assign as varchar or character, you will get the result from datastage as you instructed the datastage to do.

[kumar_s]

Not sure whats wrong in your case
Is your output and input is integer?
Is it a composite key?
Are you sure that your output is not messed up the partition provided?
For testing run you job in sequentiol mode.

[DSguru2B]

What exact syntax of sort are you using. Also give us a bit of sample data. We can only tell you whats going wrong by looking at the syntax you are providing.

[lokesh_dhiran]

I am calling the unix sort from the Datastage SORT stage. and setting an integer field as the key. Still the sorting is being done character wise. To tell you the example My key is DeptNo it values are 30,100,200,300,600
now the sorted order coming is
100
200
30
300
600

Does anyone have a reason for this. What i can make out is that it is sorting character wise from the first character

[ArndW]

Can you post your UNIX sort command line options?

[lokesh_dhiran]

hi,
am pasting the osh script generated by datastage

-- Here it goes

#################################################################
#### STAGE: Sort
## Operator
psort -sorter unix
## Operator options
-key 'DeptNo'
-asc
-key 'EmpNo'
-asc

## General options
[ident('Sort'); jobmon_ident('Sort'); seq]
## Inputs
0< 'Employees:Employees_Info.v'
## Outputs
0> [] 'Sort:After_Sort.v'
;

#################################################################
#### STAGE: Employees
## Operator
import
## Operator options
-schema record
{final_delim=end, delim='|', quote=none}
(
EmpNo:int32 {quote=none, max_width=3};
FName:string[max=255];
LName:string[max=255];
Gender:string[max=255];
DeptNo:int32 {quote=none};
Salary:int32 {quote=none};
DOB:date;
)
-keepPartitions
-rejects continue
-reportProgress yes
-firstLineColumnNames
-file '/home/kdude/DS_ex/input/Employee_sort.txt'

## General options
[ident('Employees'); jobmon_ident('Employees')]
## Outputs
0> [] 'Employees:Employees_Info.v'
;

#################################################################
#### STAGE: DS_Sort05_Output
## Operator
export
## Operator options
-schema record
{final_delim=end, delim=',', quote=none}
(
EmpNo:int32 {quote=none};
FName:string[max=255];
LName:string[max=255];
Gender:string[max=255];
DeptNo:int32 {quote=none};
Salary:int32 {quote=none};
DOB:date;
)
-file '/home/kdude/DS_ex/output_ds/Sort05_Output.txt'
-overwrite
-rejects continue

## General options
[ident('DS_Sort05_Output'); jobmon_ident('DS_Sort05_Output')]
## Inputs
0< 'Sort:After_Sort.v'
;
--End

[kumar_s]

-key 'DeptNo'
-asc
-key 'EmpNo'
-asc

But you told it was just single field...
Check with the combination of DeptNo along with DeptNo.

[ray.wurlod]

I don't see any -n options in the sort specifications.

[lokesh_dhiran]

Thats the problem.I am not able to set the parameters for the unix sort so that it sorts the field as number. It by default calls the normal sort function.

[lokesh_dhiran]

Actually i have tried with single key also. this is the osh of the job am working on. It does not matter whether you have single or two sorting keys.

[kumar_s]

Hi Lokesh,
Do you mean to say that you got the same OSH even if you have single key?
I make quick test job, sequential file--->sort---->sequential file.
I make the key as varchar and sor the similar result as you got. When i turn back to integer it gives in the order.
So Perhaps you can look on something else like, if some space or special character present in the field, which makes the sort unpredictable.
And you havnt provided the partition information yet.