I can't seem to find the documentation on the REPLACE function regarding replacing with CHR(10). When run Windows, CHR(10) ends up as CHR(13) + CHR(10). When run on UNIX, CHR(10) is just CHR(10).
The behavior seems to be non-portable. Is this expected?
Use "~n" instead. That will be OS appropriate.
That will not solve the CHR(10) replacement issue when you just want to truly replace something with CHR(10).
Meaning that you want to run on Windows and create a Unix style file?
Somewhat but the opposite (Windows style text).
I'm basically creating a normalization routine to a Windows (CR + LF) configuration regardless if the routine was run under UNIX or Windows. I can detect OPSYS to mitigate the problem. I'm just a little surprised about CHR(10) conversion because the REPLACE function documentation doesn't mention it.
Can you provide code that demonstrates what you're seeing? I can't get the same results and I don't see anything in the CHR or REPLACE functions that would cause it.
DEF VAR a AS CHAR INIT "abc".
a = REPLACE(a, "b", CHR(10)).
MESSAGE ASC(SUBSTRING(a, 1, 1)) ASC(SUBSTRING(a, 2, 1)) ASC(SUBSTRING(a, 3, 1)).
This code shows:
97 10 99
If I write the string to a file the LF is converted to CR/LF, but that conversion can be avoided with the BINARY option on the OUTPUT TO statement.
I'm writing to MEMPTR using PUT-STRING. Will that be the same as writing to a file?
I don't think PUT-STRING would cause this either. Could you demonstrate it with a short program?
My issue was coming from the OUTPUT TO.
Does COPY-LOB have the same issue?