Re: Multi-threading CRTDUPOBJ

[CRPence <CRPbottle@xxxxxxxxx>]

  Given the list of names is already in a file, then there is the 
option to let the database do the work; deciding whether to use parallel 
processing, then both to create the threads and run the work in those 
threads.  Need only to create a function that allows running parallel, 
then SELECT that function using the SQL.  Then:

       select dupDBFintoQTEMP(FileName) as Success
            , FileName
       from FileList
       where FileType='PF'

  Success could be defined as the resolved Library name from the 
completion message for the CRTDUPOBJ FROMLIB(*LIBL), and perhaps 
'*ERROR' when the request fails.  The above SELECT could be in a CTE or 
NTE from which only unsuccessful requests are included in the result set 
of the final SELECT [e.g. using where success='*ERROR', for which 
not-EOF on a FETCH means overall failure, and thus review of result set 
via that and more FETCHes is optional].  Or similarly, if just success 
versus at least one failure, a SELECT INTO or even a SET using a SELECT 
COUNT(*) as indication of effect [such that not even one FETCH is 
required].  The SELECT could be a SQL cursor from which the results are 
reviewed with FETCH; the UDF should not be on the WHERE clause.  The 
SELECT could be in a CREATE TABLE, an INSERT INTO, ??

  While I would generally use a more generic ExecCmd UDF for ad hoc 
stuff, I would probably use a function more specific to the task for 
anything else.  Thus why I did not compose a query that generated the 
full command string to effect the Create Duplicate Object request as an 
example.

Regards, Chuck

On 04 Mar 2013 05:37, RPGLIST wrote:
> I am trying to multi-thread the creation of some QTEMP objects, but
> it is not working as it should be. I did check and CRTDUPOBJ is
> thread safe.
>
> I have a file which contains a list of objects that I need to create
> in QTEMP, now I have tested this without using the multi-threading
> code and everything works fine, so it has something to do with the
> multi-threading portion.
>
> Any help would be greatly appreciated.