W dniu 24.10.2024 o 09:17, Jason Cai pisze:
Hi all
I followed these procedures:
1. Created a Variable-Length Dataset on the Mainframe
Dataset Name: TEST.DATA1
Logical Record Length (LRECL): 10
Record Format :VB
Content: The dataset contains 2 records:
000001 AAA
000002 BB
2. Downloaded the Dataset to Linux
Using an FTP on the z/OS platform, I downloaded the dataset to Linux with the
following commands:
ftp -v -n <<EOF
open [linux_server_address]
user [username] [password]
prompt
binary
locsite rdw
put TEST.DATA1
close
bye
3. Uploaded the File Back to the Mainframe from Linux
On a Linux platform, I attempted to upload the file back to the mainframe using
the following FTP commands:
ftp -v -n <<EOF
open [mainframe_address]
user [username] [password]
prompt
binary
quote site rdw recfm=vb lrecl=10
put TEST.DATA1 TEST.DATA2
close
bye
EOF
Problem Statement
I expected TEST.DATA2 to be identical to TEST.DATA1, but the RDWs were
misinterpreted, resulting in incorrect data.
How can I correctly FTP a variable-length file to the mainframe so that the
uploaded dataset matches the original?
Could you please provide guidance on the correct procedure to FTP a
variable-length file to the mainframe, ensuring that the dataset on the
mainframe is restored to its original format?
I don't know where I went wrong in step 3. Uploaded the File Back to the
Mainframe from Linux.
Any suggestions and help would be greatly appreciated
Few remarks
1. You *CANNOT* upload "variable-length file" [from Linux] to the
mainframe.
You can upload *just a file* from Linux to z/OS. A file on Linux has no
structure, it is "byte oriented", which means no records, just ordered
set of bytes. Of course we know "file formats" like XLS, or PDF, etc.
but it is up to the application to interpret the set of bytes.
2. You may want to upload text file, with some end of line control
character, like CRLF or just LF. Usually we want to change end of line
characters to records.
3. When you download text dataset to Linux/Windows you may want to
translate record boundaries to LF/CRLF characters.
4. Caution: z/OS environment (i.e. ISPF editor) tend to cut ending
spaces (X'40 EBCDIC) and then change record length. That can be a source
of discrepancies. Is it important? Usually not.
5. If you really want to transfer (only transfer, no changes!) using
Linux platform then you can simply pack your file into some envelope.
IMHO the simplest method is to use XMIT. Then you can transfer
XMIT-packed file wherever you want and finally un-XMIT (RECEIVE) the
original content. However I guess there is something to change on Linux
side. :-)
HTH
--
Radoslaw Skorupka
Lodz, Poland
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
