On 2012-06-12 07:54, Jake anderson wrote:
Thanks for your reply. I presume this would need little more research and
get back to you all once I recover it.

I have done something like this in 2005(?). The XMIT file was screwed and I could not get it back, but the contents, a PL/I library had sequence numbers and the PL/I was pretty standard, XXX: PROC ... END XXX; and it was FB(80).

I just treated it (on the PC) as a binary file, used a hex-editor to chop off the mangled XMIT headers and then read it into a text editor that could handle fixed length records. Removed the XMIT control characters, obvious from sequence numbers that were suddenly no longer aligned, saved it back as a single record and repeated this process a few more times. It took a couple of hours, but in the end I got all source back.

It will take you time, but text is recoverable.

Robert AH Prins

On Tue, Jun 12, 2012 at 12:53 PM, Paul Gilmartin <paulgboul...@aim.com>wrote:

On Mon, 11 Jun 2012 07:23:22 -0500, McKown, John wrote:

If you mean that they FTP transferred an XMIT file via an intermediate
system which was ASCII based (such as Windows) and forgot to do a BINary
transfer at some stage, you are out of luck. The problem is that, in
general, if you do an EBCDIC to ASCII to EBCDIC tranlation which include
"non printable" character, you don't get back the original file. The reason
is because multiple EBCDIC characters translate to the same ASCII
character. So there is NO way to know what the original character is.

The OP sent me, privately, a sample of his data.  Excerpts from
my analysis/reply to him:

It was apparently a PDS containing several (perhaps 5) Rexx EXECs.
Apparently a TRANSMIT OUTDATASET() was performed.  That output
data set, about 100kB, was FTPed to the PC in ASCII mode.  Is
the EBCDIC instance lost from z/OS?
Recovery is tedious.  I haven't the resources to help you further.
Some of the members (a little more than half the content) had
sequence numbered lines; those numbers could be useful markers
in reconstructing the data.

[Abby Sciuto or Penelope Garcia would do it in seconds.]

You might take this back to IBM-MAIN.  ... But, please,
answer completely any questions people ask you; don't
expect others to do all the detective work.

For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN

Reply via email to