On Tue, 27 Jul 2021 17:01:56 -0500, Frank Swarbrick wrote:
>We have a vendor that is providing a file that is EBCDIC (IBM-1140) encoded,
>but also includes an NL record/line terminator. The source system is NOT a
>mainframe system. I'm trying to figure out how to FTP the file to the
>mainframe and have it treat NL as, well, NL; i.e. a record terminator. Binary
>mode (no SITE options) doesn't work because it stores the NL characters.
>ASCII mode (no SITE options) doesn't work, I believe because it still expects
>the CRLF delimiter. I tried specifying "SITE TYPE E" (EBCDIC) and that also
>does not eliminate the NL delimiter.
>
>Any thoughts? We're seeing if the vendor can just not use a delimiter at all,
>but no luck yet.
>
Doesn't z/OS use NL as its line separator? Verify/refute this with:
echo 'foo
bar' | od -tx1
I'd expect you to see:
0000000 86 96 96 15 82 81 99 25
0000010
where the x'15' is the NL. I expect transfer in binary to preserve the NL and
simply work.
>Note: They can create it in UTF-8, but they are including the UTF-8 Byte Order
>Mark (BOM). I am able to get z/OS to strip the BOM, but I have to specify the
>transmission as being "multi-byte", so the destination has to be VB. Which we
>can deal with, but we'd prefer FB as that is how we have it from the old
>vendor.
>
Use of a BOM with UTF-8 is generally deprecated.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
