This thread looks to have played out, but I wanted to add that I would not love 
raising an error on a MOVE. It just feels like that's at the wrong level; it's 
not a data exception per se (well, I guess it COULD force one, but that would 
be kinda bogus) and so what is it? In other words, here's a program that works 
99% of the time, but fails occasionally because you're throwing user input at 
it and hey, somebody managed to mix characters. That's going to be a nightmare 
to document, and users are going to get it wrong.

Far better IMHO to force the user to call a conversion function which itself 
has documented failure modes, which they'll (hopefully) test for (and if they 
don't, then it's on them).

Not sure I'm getting my point across here; it's sort of like "This instruction 
works except on alternate Thursdays"--nobody is going to grok that properly and 
handle it cleanly.

Sounds like you have a solution, anyway; just had to get this in there. 
Unicode/UTF-n are amazing achievements, but going between UTF-n and EBCDIC is 
always fraught!

-----Original Message-----
From: IBM Mainframe Discussion List <[email protected]> On Behalf Of 
Farley, Peter
Sent: Friday, January 2, 2026 6:59 PM
To: [email protected]
Subject: Re: Please vote for COBOL "idea" for reverse character translation

I personally would be satisfied for the runtime logic to just raise an LE error 
condition and dump for any sort of conversion-stopping error.  So long as 
normal “character not defined in the requested EBCDIC CCSID so make it SUB in 
the output” issues are just done without asking, I am perfectly willing to scan 
the result for SUB characters myself and use application-dependent error logic 
to handle or ignore the result as needed.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to