I have a suite of tools for processing IBM COBOL source code; these tools are built as Win32 applications and talk to Windows (including network) files using traditional Windows file system calls (open, close, read, write) and work just fine, thank you.

I'd like to integrate these with Eclipse; we understand how to get Eclipse to do UI for us we think. The problem is that Eclipse/RDZ users access mainframe files through some IBM magic. In How does RDZ access mainframe files I tried to understand how Eclipse accessed files on a mainframe. Apparantly Eclipse/RDZ has a secret filesystem access backdoor not available to normal mortals.

At issue is how our tools, reading some Windows-accessible file (local disk file, NFS to mainframe, ...) can associate such files with the files that Eclipse can access or is using? Ideally we'd like UI-integrated versions of our tools take an Eclipse file-name string for a mainframe file, pass it to our Windows application to process, have the Windows application open/read/process the file, and return results associated with that file to the Eclipse UI.

Is there a canonical file name path that would be used with mainframe NFS that would be equivalent to the name or access object the Eclipse RDZ used to access the same file? Are all operations doable internally by Eclipse, doable by the mainframe NFS [for instance, can NFS read/update an element in a partitioned data set? Can Eclipse RDZ? Does it matter?]

Is the mainframe file access available to custom Java code running under Eclipse RDZ (e.g., equivalents of open/close/read/write based on filename/path/something?) If so, can somebody steer me towards documentation describing the access methods?

Anybody else already solve this problem or have a good suggestion?