Ref: Your note of Thu, 19 Oct 2017 12:13:48 -0400 z/OS Unicode Conversion Services are only available on z/OS. There's a clue in the name. ;-)
TROO is a z instruction, which is not available in ESA/390 mode when running on CMS, which HLASM still supports. As far as I know, there's no fast interface to iconv for translating data in storage. I use it on VM for converting data files between UTF-8 and EBCDIC (DBCS if necessary) after using FTP to fetch data in binary. SUPERC sets up an ASCII translate table in its common storage area during initialization by copying a 256-byte base table and optionally applying a short list of modifications if some other table is being used. At run-time, ASCII translation is just a translate instruction, so the method of definition has no impact on performance. One apparent bit of good news: Although comments in the HLASM source say that the ASCII translation for CA fields is supposed to leave bytes unchanged if they do not map to 7-bit ASCII, I cannot find any reference to this in the current publications, and indeed it specifically says that it maps 37 to 819, so I think we are free to extend that table (used for CA) to include the whole 256-byte mapping of code page 37 to 819. I have also checked that the existing mappings are a true subset of the standard full mapping of code page 37 to 819, for example as in the ASMALTAS table. Another bit of good news is that it turned out to be easy to add support for CU self-defining terms, so it should be possible to provide the complete set (CA, CE and CU). Jonathan Scott HLASM team, IBM Hursley, UK
