Patrick O'Keefe wrote:
On Sat, 1 Dec 2007 11:07:06 -0600, Joel C. Ewing
<[EMAIL PROTECTED]> wrote:
..
I notice your table doesn't mention the "¬" encoding differences
between
IBM-1140 and IBM-1047, but perhaps that character wasn't relevant
in the
context of that discussion.
...
Once again, the character seen in your post depends on the
codepage used to display it. I see a "logical not" symbol, and
assume that is the character you are refering to.
I found lots of references to the IBM-1140 codepage on web but
have not seen its contents, so my following comment may be way
off base.
I think the problem Steve was mentioning was when a character has
different codepoint in different codepages.
Does IBM-1140 havea "not" as something ther than x'5F'?
Most of the codepages I've looked at (which admitted is not many)
either have the "not" symbol as x'5F' or don't have it at all. Most
codepages I've used lately have the "caret" for x'5F' and don't have
a "not" symbol at all. I've taken to interpretting "caret" as "not".
The only places I've seen the "not" sysmbol used are PL/I and REXX
(although I suspect there are many others, too). I haven't had
easy access to PL/I since 1988 so I'm absolutely out of date.
As far as REXX is concerned there are several alternatives to
"not" - slash-equal and backslash-equal instead of not-equal.
In fact, REGINA uses backslash-equal and caret-equal ... assuming
I'm using the right codepage to look at the REGINA doc. :-)
Pat O'Keefe
...
I also have been unable so far to find any on-line reference that
actually gives examples or names of established "standard" graphic
representations for the various codepoints in the EBCDIC codesets. I've
have been running of necessity under the assumption that there is some
standard and that IBM's PComm emulator would conform to it, but that is
probably a rash assumption. It's also quite possible (likely?) that
whatever standard exists may have left some codepoints undefined, and
that PComm could have implemented graphics for those codepoints that
differ from other 3270 emulators.
I have done some tests to determine the graphic symbols displayed in my
own environment with IBM PComm Ver 5.6 sessions set to codesets
IBM-037-US, IBM-1047-US, IBM-1140-US, and IBM 924-Multinational and have
posted the results (as images to avoid issues with the displayer's text
codeset) at http://home.earthlink.net/~jcewing/Codesets/
This shows that by PComm's implementation of the codesets that, among
other differences, there is an interchange in the graphics for PL/1
Logical not and the Circumflex or caret at codepoints X'5F' and X'B0'
between IBM-037-US and IBM-1047-US. It also shows that the differences
between IBM-924-Multinational and IBM-1047 are much more extensive than
just the Euro symbol, probably because there was no US-specific IBM-924
variant (at least not in PComm 5.6).
These display images were created from a 3270 session at the TSO Ready
prompt to avoid any issues with unexpected translations by ISPF, using a
quick and dirty REXX exec that simply displays 16 text strings
containing the hex character values from x'4x' through x'Fx' for x=
0,...,F. At some point I may try similar tests using ISPF panels, but
the combinations of interest become much larger once you add the
complexity of ISPF panel CCSID and ISPF terminal type translations to
the picture.
I believe REXX allows the slash alternative operators precisely because
translation with codesets like IBM-037 or IBM-1140 can cause syntax
problems and algorithm failure if the caret is used.
--
Joel C. Ewing, Fort Smith, AR [EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
