encoding problem with EBCDIC/ASCII

[Lu Battist]

I'm creating a file that has to be '|' (pipe) delimited from a java program running on unix (ASCII). It then get transfered to an ibm mainframe (EBCDIC) through a secure ftp type of program that I don't have control over. From there it gets transfered over the net and in the hands of the customer, probably back into a unix file again.

That pipe, on the customer end, shows up as a superscript 3 character (as in x cubed). On my end from the unix side, the pipe is a hex 7C. I have no idea what the hex code for the superscript 3 is on the customer end. No other characters seem to be affected, probably because they are standard ASCII (1-127) characters that translate well.

Anybody know a way to get the correct character to the customer end?
Even knowing which encoding produces a superscript 3 might be useful as it doesn't seem to be on any EBCIDIC/ASCII map I've seen.

Thanks.

[Paul Clapham]

That's Unicode you have in your Java program, actually, but since the character is 7C then it's in the Basic Latin block that's a copy of ASCII. And it represents a vertical bar character, which you probably already knew that.

So first it goes through FTP, which is supposed to convert it to EBCDIC. From what you say that's working pretty well. And then it goes through some other conversions that map it back to Unicode in a web page. Do you have any control over those conversions? If so, you will need to make sure the correct encoding is being applied in the conversion back from EBCDIC.

There are various flavours of EBCDIC and they don't all use the same code point for the vertical bar character, so make sure everything is configured correctly and knows which EBCDIC flavour it is working with.