MVS File System

The file system is essentially flat, i.e. there are no file trees. There are so-called Partitioned Data Sets (PDS) resembling the VMS text libraries. Typical MVS dataset name look like these:

'KP09.LMDV.RUN0000'      # a Sequential Dataset with Raw data on the FOPI account
'KP09.FLIB.FORT(PAWFOPI)' # a PDS member for a FORTRAN source file
PAW9102.FORT(EUSER)       # a private EUSER routine source member

Names enclosed in single quotes are considered "fully qualified". Other dataset names are completed by MVS with your user prefix, which normally equals your User ID (=account). This mechanism resembles the "default directory" on VMS. You can switch to another users "file system" with the TSO command

prof pref(kp09)           # example to switch to the FOPI account

Be careful, especially when switching to KP09 !! You may destroy valuable data when you issue dangerous commands. Dataset names can have up to 44 characters (A-Z,0-9,$,#,@, not the underscore,_ ). Name parts can be separated by as many dots as you like, provided the resulting name is not longer than 44 characters. Name parts as well as PDS member names can have up to 8 characters, the leading one must be non-numeric. Space for user data (program sources etc.) is allocated from a disk pool (USERxx) shared by all users, there are no disks dedicated to specific users or user groups like on VMS or UNIX. The quota for user data is normally 100MB ( 200000 VMS blocks). You can recover "lost" (accidentally deleted or changed) datasets for example with

hrecover flib.fort NEWNAME(flibrcov.fort) DATE(mm/dd/yy)

to get back an older version.