Re: BLOCK(*YES) questions

[Simon Coulter <shc@xxxxxxxxxxxxxxxxx>]

On Sunday, February 29, 2004, at 11:57  AM, M. Lazarus wrote:

>  Are you referring to the OVRDBF NBRRCDS() keyword for double 
> buffering?  I thought that double buffering in RPG only worked for 
> PRINTER files and that it didn't really work anymore.

Yes. NBRRCDS controls blocking from DASD into main storage. 
SEQONLY(*YES nnn) controls blocking from main storage into the program 
buffer. Blocking used to work only for sequential processing because 
you couldn't block records for files defined as keyed in the program. 
As soon as you put a K on the F-spec program buffer blocking was 
disabled. BLOCK(*YES) changes that behaviour and blocks both sequential 
and keyed files but this happens after records are brought from DASD to 
main storage.

I'm not sure if BLOCK(*YES) affects blocking into main storage or only 
program buffer blocking. I suspect the latter. Barbara or Hans could 
confirm this.

NBRRCDS is useful only when the records on DASD are physically in the 
sequence they are processed so using NBRRCDS in conjunction with 
BLOCK(*YES) will help only for files processed sequentially or 
sequentially by key.

NBRRCDS should not be used for files where the physical order of the 
records is not the same as the order in which they will be processed 
because that will cause additional, possibly unwanted, records to be 
brought into main storage.

A better way to improve performance for this sort of random processing 
is to use SETOBJACC to load the file into a dedicated storage pool.

Regards,
Simon Coulter.
--------------------------------------------------------------------
   FlyByNight Software         AS/400 Technical Specialists

   http://www.flybynight.com.au/
   Phone: +61 3 9419 0175   Mobile: +61 0411 091 400        /"\
   Fax:   +61 3 9419 0175                                   \ /
                                                             X
                 ASCII Ribbon campaign against HTML E-Mail  / \
--------------------------------------------------------------------