On Mon, 5 Aug 2013 10:34:56 -0400, Shmuel Metz (Seymour J.) wrote: > >The problem is Unix, not EBCDIC. The ASCII LF is *not* a new line >function, but half of a new line function. The DOS CRLF convention >reflects ASCII. '15'x is a correct translation of the Unix new line >but not of the ASCII LF. '25' is a correct translation of the ASCII LF >but not of the Unix new line. So in generally you need to know whether >you are dealing with real ASCII or with the Unix variation in order to >translate properly. > All of which is comparable to, and irrelevant as, quibbling about whether, in EBCDIC, x'F1' is a "real" representation of the CCW to skip to top-of-form.
The problem arises because while Linux iconv(1) and CMS PIPE TRANSLATE convert between ISO-8859-1 and IBM-1047 according to the published specification of the code pages, z/OS iconv(1) violates IBM's own specification. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
