2

I have an application on z/OS USS that happily reads EBCDIC (IBM-1047) and ASCII (ISO8859-1) files that are tagged with either encoding into char[] buffers. When started from a shell, the C runtime will automatically convert the file contents in fgets() into EBCDIC for the program to use. This allows comparisons with literals encoded natively in EBCDIC.

However, when the program is run via Jenkins (Java JVM 1.8), the automatic conversion is suppressed and fgets() returns data unchanged, so ASCII. I have tried supplying every environment variable in the user's .profile and /etc/profile to the JVM, and verified that they are in the program's envp.

Does anyone know where the control of this function is documented or available?

VillageTech
  • 1,968
  • 8
  • 18
Randall Becker
  • 65
  • 3
  • Can you post your locale variables, ex. `LC_COLLATE` or preferably all? From the "application on z/OS" and from jenking environment, if somehow available? What do you use for string comparisons? – KamilCuk Jan 16 '20 at 20:27

1 Answers1

2

One of the ways of controlling automatic conversion is _BPXK_AUTOCVT=ON This article here describes the issue on more detail.

Here is a snippet:

enter image description here

enter image description here

Hogstrom
  • 3,581
  • 2
  • 9
  • 25