Re: Count with no Data and no Key ? (CKD internals again)

1. A BDW with a length field of 4 would give you a DASD record with KL=0, DL=4, which does not generate a UE. 2. I haven't looked at the code for anything newer than 3420 GCR (6250 BPI), and have never tried it on 3480 or later, but I would expect wonky results for short blocks. I

Re: Count with no Data and no Key ? (CKD internals again)

It's actually the DASD controller that detects EOF on a Read Data CCW. As for PDSE, the dataset is formatted into control intervals and there are data in the dataset used by the access method to delineate logical records and members. -- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3

Re: Count with no Data and no Key ? (CKD internals again)

I've done the same thing. In my case, it was TSOE TRANSMIT, which at the time used VIO for the temporary data set created for transmission purposes. That can be changed in PARMLIB IKJTSOxx TRANSREC to something other than VIO. VIO really makes no sense in a modern mainframe environment but can

Re: Count with no Data and no Key ? (CKD internals again)

On Fri, 23 Feb 2018 14:49:05 +, Christopher Y. Blaicher wrote: > >The access methods detect the key=0, count=0 records and process them as EOF, >which is how they signify the end of a PDS member. Every member has an EOF >record with the rare exception of a member

Re: Count with no Data and no Key ? (CKD internals again)

Radoslaw, The access methods detect the key=0, count=0 records and process them as EOF, which is how they signify the end of a PDS member. Every member has an EOF record with the rare exception of a member that ends at the end of the track on the last track of a data set. I don't know how

Re: Count with no Data and no Key ? (CKD internals again)

I don't know internals, but definitely one can use short blocks on DASD, 18B is not a limit. You can use 18 or 5 if you want. Of course the tracks utilization will be extremely poor, but that's different story. BTW: I'mt not sure, but it seems the null record inside a member is when the

Re: Count with no Data and no Key ? (CKD internals again)

Paul Gilmartin wrote: >Also, just curious, I understand that on tape a block <18 (? still?) is >bypassed as a noise record. Wow! Interesting! Where is that documented? >But it's possible that with RECFM=VB the last block might, by happenstance, >consist of as few as 8 bytes (BDW=8; RDW=4).

Re: Count with no Data and no Key ? (CKD internals again)

On Thu, 22 Feb 2018 15:55:49 +, Seymour J Metz wrote: >I seriously doubt that; it would generate a Unit Exception trying to read the >second record of the member. You're clearly misinterpreting something. > >There is a EOF after the directory and an EOF after each member. There is no >EOF

Re: Count with no Data and no Key ? (CKD internals again)

I seriously doubt that; it would generate a Unit Exception trying to read the second record of the member. You're clearly misinterpreting something. There is a EOF after the directory and an EOF after each member. There is no EOF in the middle of a member. -- Shmuel (Seymour J.) Metz

Re: Count with no Data and no Key ? (CKD internals again)

I just found that in PDS every (?) data record is interleaved with null record (count, no key, no data). More precisely: null record is at the end of directory and between all member records, even for members occupying several records. So PDS contains multiple EOFs, even more than one EOF per

Re: Count with no Data and no Key ? (CKD internals again)

Zero data count is an EOF regardless of the key length. -- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 From: IBM Mainframe Discussion List on behalf of Christopher Y. Blaicher Sent:

Re: Count with no Data and no Key ? (CKD internals again)

You can have a zero length key and data, it is called an End Of File (EOF) marker. Chris Blaicher Technical Architect Mainframe Development P: 201-930-8234 | M: 512-627-3803 E: cblaic...@syncsort.com Syncsort Incorporated 2 Blue Hill Plaza #1563 Pearl River, NY 10965 www.syncsort.com Data