Ref:  Your note of Thu, 19 Oct 2017 11:18:43 -0600

On Linux, HLASM looks at the first record of a file and if it
contains an ASCII space, asterisk or digit 0-9 (all of which are
outside the EBCDIC printable range) it assumes the file is ASCII
and turns on translation from 819 to 037.

We are already aware of a requirement to be able to translate
SYSPRINT to ASCII for Linux. I'm not sure about translating to a
general codepage, as I would have thought that one would normally
view the listing using the same code page as the source.

Paul Gilmartin writes:
> I think it would be best for HLASM to operate with a consistent internal
> CP regardless of the input CP(s).  "DC C'['" should generate the same
> binary object regardless whether the input CP is 819, 1047, 500, or 037.

That seems a bit backwards.  HLASM reads files containing EBCDIC
codes which it normally regurgitates faithfully, unchanged, in
character constants, without any idea what symbols they
represent.  The same code may appear as different characters on
different terminals or printers, especially when it is supposed
to represent a square bracket!  In order to translate that code
reliably to a representation of the same character in another
code page, it needs to know the input code page.  You can of
course use the TRANSLATE option with a user-defined table to
translate character strings in the object code to a different
EBCDIC variant if necessary.

Jonathan Scott
HLASM team, IBM Hursley, UK

Reply via email to