On 2016-04-13 19:15, Steve Thompson wrote:
> Gil, not true.
> 
> RECFM=U was used by OBS/ACS WYLBUR to be as efficient as possible with disk. 
> So unless told otherwise, WYLBUR would save your source or data in RECFM=U 
> files. Further, it would do compaction when it ran into an out of space 
> problem, and it would do it so that the DSN in question would be a single 
> extent at the end of the process.
> 
> I do not know if ROSCOE or EDGAR (I think it was a different product whose 
> name I can't remember right now) also did RECFM=U or not, for the same 
> reasons.
> 
> If the DSN were RECFM=FB (for instance), then WYLBUR would honor that.
> 
> Don't ask me all the particulars of how WYLBUR did all this because I haven't 
> touched WYLBUR internals since about 2000.
>  
IOW, you don't know whether WYLBUR (or ROSCOE or EDGAR) supports PDSE.  Any
of them may have been updated.

> On 04/13/2016 08:14 PM, Paul Gilmartin wrote:
>>
>> Customarily, RECFM=U indicates a Program Object.  Perhaps PDSE enforces this.
>>
I said "Customarily", not "unconditionally".  And I believe ISPF enforces its
own restrictions.  Will ISPF Edit a RECFM=U member?


On 2016-04-13 18:56, Savor, Thomas (Alpharetta) wrote:
>
> You do receive an EODAD on the next read, but in my example of 500 lines in a 
> program....
> first Read = full block (349 records), second Read = short block (actually 
> 151 records, but program thinks its 349), next read gets EODAD.
>On 2016-04-13 18:56, Savor, Thomas (Alpharetta) wrote:
>
> You do receive an EODAD on the next read, but in my example of 500 lines in a 
> program....
> first Read = full block (349 records), second Read = short block (actually 
> 151 records, but program thinks its 349), next read gets EODAD.
>
So your 151 records occupy 12080 bytes. What appears in the remaining 15840?
(Probably unpredictable; don't count on it.)

Note that a short block may appear in the interior of a file.  Unlikely for a
PDS member, but you should be prepared to handle it.

>> Use QSAM with LRECL=80.
>
> I'm using BPAM, LRECL=80, DSORG=PO.
>
There's a hack involving zapping the JFCB.

> I'm processing all members of a PDS or PDSe. Could be 5 members or 5000 
> members, either way program 
>
I would expect that for compatibility PDSE supports BPAM *provided* you don't
lie about RECFM and LRECL.  How does BPAM report short blocks?  I recall
something about inspecting the RBC in the IOB.

-- gil

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

Reply via email to