::cobol
IDENTIFICATION DIVISION.
PROGRAM-ID. 'TESTME'.
ENVIRONMENT DIVISION.
CONFIGURATION SECTION.
SOURCE-COMPUTER. GC-MACHINE.
OBJECT-COMPUTER. GC-MACHINE,
PROGRAM COLLATING SEQUENCE IS TESTME.
SPECIAL-NAMES. ALPHABET TESTME
'A' also 'B' also 'C' also 'd' also 'e' ALSO 'f',
'g' also 'G'.
DATA DIVISION.
PROCEDURE DIVISION.
display "#" function ord('G') "#" end-display
display "#" function char(1) "#" end-display
if 'A' = 'f'
DISPLAY "Yes it is"
end-if
.
gives
::text
#000000072#
##
Yes it is
(I removed the ASCII 01 that displays on the second line, just to avoid weirdness in the ticket)
That 00000072 is not right (or my understanding of the standard is not right; more than willing to believe that could be the case. Perhaps a missed step in my local setup?).
It seems the intrinsics are falling back to native alphabets, while the compare is honouring the COLLATING SEQUENCE.