Gilbert, I move AFP files all the time. Of the 3 methods mentioned in your reference, http://www-01.ibm.com/support/docview.wss?uid=swg21188301 , I use Method 3, tersing, the most. It seems the easiest and most foolproof to me. You just terse the original file, then make sure all transfers are done in binary mode, not text, and unterse the file on the new z/OS system. For the final FTP upload, I specify something like SITE RECFM=FB LRECL=1024 BLKSIZE=0 CYLINDERS PRIMARY=10 SECONDARY=10 where you adjust the size to fit your requirements. Then I do my binary FTP upload to z/OS. Untersing the data set restores the original DCB attributes.
AFRREBLK is for fixing untersed binary AFP files that have been uploaded from a non-MVS system. If you have PSF, you should find SYS1.SAPSEXEC(AFRREBLK) and the sample profile SYS1.SAMPLIB(AFRREBLK). Read the comments inside the exec for usage details. Text files are a little more forgiving, but I occasionally get one somebody has sent to me that is all one long wrapped file. Chances are it has CRLF type characters in it. One way to fix some of those files is to use the OMVS system to split it. I have a little exec like this: BROWSE BOLAN.MY.CLIST(SPLIT) - 01.02 Line 00000000 Col 0 Command ===> Scroll === ********************************* Top of Data *************************** /* REXX COMMAND TO FIX UNSPLIT HFS FILES */ /* Roger Bolan May 18, 2010 */ Trace C ADDRESS tso PARSE ARG olddsn if olddsn = '?' then call explain stripdsn = Strip(olddsn,'B',"'") "OPUT '"||stripdsn||"' '/tmp/"||stripdsn||"'" "OGET '/tmp/"||stripdsn||"' '"||stripdsn||"'" "TEXT" "OSHELL rm /tmp/"||stripdsn return RC explain: say 'Use from ISPF 3.4' say 'Example format: %split / ' say 'use / to enter the quoted fully qualified name from ISPF 3.4' exit 0 Sometimes text files have only a x'15' or x'25' byte dividing the original records. I have a VM system too, so I use pipe commands, like the following excerpts, sometimes. /* splits on 0x15 */ 'PIPE <' fn ft fm '| DEBLOCK LINEEND | >' fn ft 'A' /* splits on x'25'*/ 'PIPE <' fn ft fm '| DEBLOCK C | >' fn 'SPLIT A' Regards, --Roger On Thu, Jan 10, 2013 at 10:12 AM, retired mainframer < [email protected]> wrote: > You have received a couple of ideas to solve the transfer problem. It's > your last sentence I'm concerned about. If the minor tweaks are strictly > limited to a character by character substitution (one for one) it may be OK > (obviously not if the dataset contains a checksum). But if any > substitution > changes the number of characters in a "record", it will mess up > XMIT/RECEIVE, TERSE, and/or the BDW/RDW to render the data unreadable by > normal z/OS I/O. > > And then there is my paranoid concern regarding why someone is altering a > mainframe dataset to make it look like the mainframe has the wrong data. > Data received from a Windows system should be easily identified as such so > problems with the data can be addressed on the system that created it. > > :>: -----Original Message----- > :>: From: IBM Mainframe Discussion List [mailto:[email protected]] > On > :>: Behalf Of Gilbert Cardenas > :>: Sent: Thursday, January 10, 2013 6:11 AM > :>: To: [email protected] > :>: Subject: FTP variable block dataset from z/OS to Windows and back > :>: > :>: Does anyone know if the following limitation still applies? > :>: http://www-01.ibm.com/support/docview.wss?uid=swg21188301 > :>: > :>: I have a request to ftp a z/OS variable blocked dataset to a windows > :>: server and then be able to ftp that dataset back to a z/OS system. I > am > :>: not privy to all the details about why they want to do this, just been > :>: asked if it is possible to do? I'm guessing they want to make some > :>: minor tweaks to the dataset and re-upload. > :>: > :>: Anyhow, can this be done and if so, can you share any samples? > :>: > :>: ---------------------------------------------------------------------- > :>: For IBM-MAIN subscribe / signoff / archive access instructions, > :>: send email to [email protected] with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
