It's been a while since I've used TSO Pipes, but as I recall you do need to 
allocate before writing, even with pipes.

IIRC, I wrote an output filter that would allocate the parameter dataset if it 
didn't exist.  It applied useful defaults (e.g. CYL SP(10 10) RELEASE RECFM(VB) 
LRECL(8192) and some SMS stuff I don't remember), generated a DD name, and then 
CALLPIPEd the input stream to the output stage.  I believe just about any 
default could be overridden, so it was fairly easy to use.  The few remaining 
cases where the preallocate would work, I bit the bullet and coded my own.  
That was rare and tolerable, since most cases had been covered already.

John, have you considered making something like this part of the TSO service?

PDSs also help a lot; they have a CMS flavor under TSO pipes.  And you only 
allocate them once.  PDSEs are even better.  Use a ddname that matches what you 
would have used for the filetype if you were working under CMS.  The member 
name is then the 'filename'.  You also get some portability with PDSs if you do 
this.  (Many kudos to John on this smart move.)

I hope this helps.

-----Original Message-----
From: CMSTSO Pipelines Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of 
Glenn Knickerbocker
Sent: Thursday, April 03, 2008 10:03 PM
To: [email protected]
Subject: [CMS-PIPELINES] novice BatchPipes data set question

OK, is there a stage that will allocate and write a data set all in one
shot?  This pre-allocating business is getting awfully tedious.

¬R

Reply via email to