The transfer of path names between the partner systems takes place in UTF-8 in character mode so that characters outside the scope previously supported by openFT can also be transferred and thus exotic path names can also be addressed.
Displaying remote file names
If a command for the reading of file information is executed in character mode, the file names obtained from the partner system are then also converted from the character code of the partner to the local character code. Files, whose names do not correspond to a valid coding according to the setting in the partner system, are skipped when listing directories. For example, file names from a Unix system are not shown with a set UTF-8 (ftmodo -fnccs=utf8) when they do not correspond to a valid UTF-8 coding. In transparent mode on the other hand file names are transformed according to the previous regulations of openFT and according to the operating systems involved and are retained as byte strings. File names on Windows, which do not correspond to the character set of the set ANSI code, are skipped.
Note on Unix systems
The character code of a Unix partner system is determined on the inbound side from the new openFT option FNCCS, which the openFT administrator of this system sets via operating parameter (e.g. using the command ftmodo -fnccs). For this purpose he should specify the predominant local coding, e.g. UTF8 or ISO8859-1. On the outbound side the character presentation for the specification of remote file names, pre-processing commands, etc. is derived from "LOCALE" and the environment variable LANG (e.g. ISO8859F for de_DE@euro, ISO88597 for el_GR, or UTF8 for de_DE.utf8). If commands are entered via the terminal or console, the terminal display should also correspond to this setting. The code table derived from the "LOCALE" and LANG variables must be integrated in openFT. Character codes, in which the coding of the ISO646 character set differs from the ISO8859-1 coding (e.g. ISO646de, EDF041, UTF16), should be avoided both for the definition of FNCCS and the LANG setting. Otherwise, pre-processing, ftexec and ftadm, or even normal file accesses would not work correctly in character mode.
Note on Windows systems
In the current Windows systems the path names are always stored in the file system in UCS-2 (a subset of UTF-16) so that no further openFT option is necessary here. Path names are coded in UTF-8 in the request description, and in character mode also for transfer.
Note on BS2000 and z/OS systems
Fixed character codes are also assumed (EBCDIC DF.04-1 and IBM1047) in character mode in BS2000 and z/OS.