The JCL manual under RECFM says spanned:   VS indicates that the records are 
variable length and spanned. VBS indicates that the records are variable 
length, blocked, and spanned, and that the problem program must block and 
segment the records.

Apparently spanned records is not part of the access method because the manual 
says the problem program must block/segment records. As you say, many of the 
IBM utilities can handle VBS because they built it in.

The access method does know about track size but I don't think any of the 
access methods controls the number of records in a block. Is there an access 
method that creates short physical records just to fill in the free space on a 
track?

It's not a problem to create a program that creates short physical records 
(using TRUNC macro). You can obtain track size and monitor your buffer usage to 
fill in the unused portion of the track if desired. I suspect this is what the 
binder does. 

Jon Perryman.


----- Original Message -----
> From: Paul Gilmartin <[email protected]>
> 
> On Sat, 7 Dec 2013 10:01:14 -0800, Jon Perryman wrote:
> 
>> FBS and VBS work the same as FB and VB except for records spanning into the 
> next block. They all ignore track size.
>> 
> The S in FBS denotes Standard, not Spanned.  In an FBS data set, all blocks
> except possibly the last are identical in size.  There is a needless 
> constraint
> that you are not allowed to MOD to an FBS data set.  Inadequate access
> methods.  UNIX uses FBA with standard block sizes, and append works fine.
> 
> How can an access method "ignore" track size?  An attempt to write a
> block larger than (the space remaing on) a track is bound to fail.
> 
>> In all my years, I've only seen VBS used once (I think it is SMF data). 
> The biggest advantage for VBS was that the size for all physical records 
> except 
> the last would be the blocksize. If you specified blocksize 32760, then each 
> track would contain one physical record of 32760 bytes and the rest of the 
> track. To make it optimal, you specified a blocksize of half a track.
>> 
> I believe IEBCOPY uses VBS.  And I believe one of the motivations for VBS
> was the requirements of FORTRAN, which wrote logical records larger than
> DASD tracks of its era.
> 
> Binder is able to use BLKSIZE=32760 efficiently.


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to