[icu-support] support for non-IBM stateful EBCDIC charsets

[Mark S. <msa...@us...>]

Hi,

We have a Japanese customer that requires support for data encoded in one
of several stateful EBCDIC charsets created by mainframe manufacturers
other than IBM.  These encodings use shift characters to bracket sequences
of double-byte characters embedded within a stream of "standard" ebcdic
characters, but the SO/SI pair is something other than the 0x0E/0x0F
currently supported by ICU:

   (manufacturer, encoding, SO, SI):
   ------------------------------------------------
   (Hitachi, KEIS, 0x0A42, 0x0A41)
   (Fujitsu, JEF, 0x28 or 0x38, 0x29)
   (NEC, JIPS, 0x1A70, 0x1A71)
   (IBM, IBM/ISO 2022, 0x0E, 0x0F)  ; already supported by ICU

Can support for the non-IBM shift sequences be added to a future version of
ICU?  I think we can get bulk data from our customer for at least some of
the conversion maps.

Regards,

Mark