The DOS hierarchical filesystem
was originally based on the UNIX system filesystem,
but there are a number of important differences:
Physical disk boundaries are hidden on UNIX systems.
Under DOS, separate physical disks can
be referenced by disk drive letter (C:, D:, and so forth).
Under the UNIX system,
disks are mounted onto one hierarchical filesystem.
A limited form of this feature is available
under DOS with the join command.
The naming restrictions are different.
/ is the directory separator for the UNIX system,
not \ as in DOS.
The UNIX system does not enforce the fixed
eight-character name, three-letter extension
UNIX filenames can be up to 255 characters long.
Note that the SCO® XENIX® system
and versions of the SCO UNIX System before Version 4.0
have a 14-character filename limit.
Filenames are case significant on UNIX systems.
In other words, ``File'' is not the same
There is no distinction between
ordinary files and devices on UNIX systems.
All devices are accessible through the filesystem.
When opening a file on the DOS system,
the C program must indicate
whether it is a text file or a data file.
If it is a text file,
a CR-LF sequence is translated into
a \n sequence for the C program
when it is read from the file.
This can cause problems in file size determination
and in seek operations on the DOS system
but is not a problem on the UNIX system,
where the line termination character is \n.
A directory entry under the UNIX system
points to a control block called an inode
rather than pointing directly to the data block on disk.
The inode then points to the disk block actually used.
This allows directory entries in different directories
to point to the same physical data on disk,
often saving considerable space.
On S5 and EAFS filesystems,
keeping files and directories small
provides significantly better file access performance.
Files that are larger than 256KB
and directories that are larger than 64MB
will slow file access times.
© 2005 The SCO Group, Inc. All rights reserved.