At 10:13 -0500 on 06/16/2015, Tony's Outlook via Mozilla wrote about
Re: RPG for the 360/20:
Oh it definitely existed, this being my first programming language,
and sorely missed. As an "intellectual" exercise I once wrote a
program to convert/interpret incoming card data that was mistakenly
loaded 12 edge face down.
I know the challenge of writing these types of odd ball programs.
I once had to write one for real. We had a 2540 Reader/Punch which
was replaced with a 2501 Reader. The problem was that we had job that
read cards that were partly Column Binary and partly EBCDIC. With the
2540 we could read the card (but not eject it) in Column Binary mode
and then read it again as EBCDIC (from the buffer in the reader)
saying to ignore the coding errors and eject the card. This worked
until the 2540 was replaced with the 2501. The problem was that on
the 2501 you got one crack at the card since it was not buffered.
Thus we read the card as Column Binary and needed some way to convert
the Column Binary image of the EBCDIC area back into EBCDIC. It
turned out to not be that hard once I though about it. It involved a
number of TRs (one used as patterned move to separate the top rows
(12-3) from the bottom rows (4-9) and others to convert 12, 11, 0, 8,
9 punches into 5 bits of a byte and 1-7 punches into a 3 bit number.
Doing an OC of the translated high rows onto the translated low rows
followed by a final TR to convert from Card Image EBCDIC to internal
EBCDIC finished the process.
As a sanity check, the same process was done with a set of TR Tables
that mapped the 1-7 rows into the low 7 bits of a byte followed by a
TRT to verify that each of these bytes had only one or no bits set
(ie: That the EBCDIC Data read as Column Binary was valid punch
combinations).
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
