The main purpose of scconfig is to detect system specific settings and
store them in a comfortable way for later processing. Each setting has
an unique name and a value. The storage mimics standard UNIX file systems
to some extents: names of the settings are like file names and "file content"
is the actual data associated with the setting. This allows scconfig to
organize settings in a tree, where "directory" names are separated by dashes.

However, this is not a true tree storage, rather a list of hash tables. Each
toplevel directory has its own hash table that stores all further subdirectories
and files. The root directory is /. There is no setting stored outside of this
tree structure. Thus a unique setting name /arg/fstools/cp (path) is split up to
"/arg/" (first level directory) and "fstools/cp" (filename). If there's a hash
defined for the given first level directory, filename is looked up there.

A special feature is linking first level directories so that "/foo/" and "/bar/"
would point to the same hash table. This means accessing filenames under any
of the linked first level directories would result in addressing the same
file. This feature is used when the host system is the target system to ensure
that /host/ is the same as /target/ without the hassle of duplicating data
and maintaining sync.

There is a "CWD" defined, if a path doesn't start with slash, it is appended
to CWD and the resulting path will be used as the full path. CWD must start
with slash but may contain more than one slashes.