RFS vs. NFS

[John Chambers]

> Even under RFS, exec'ing a VAX executeable on a Sun doesn't work very
> well, so it can matter whether the file is local or remote.  RFS thus
> does not support full UNIX file system semantics according to the
> definition ekrell gave.

Hey, wait a minute, exec isn't really a file-system operation, except
in the trivial sense that everything is a file-system operation.  (After
all, /dev/mem exists, so a store operation changes a file...:-)  File
operations are things like open(), read(), .... 
 
> I have my doubts whether there is any adequate definition of UNIX file
> system semantics for a heterogeneous distributed environment.

Well, an interesting argument is that most Unix systems have file
systems made up of multiple disks and/or partitions mounted together.
The only thing new in a 'distributed' system is that the disks or
partitions are owned by different processors.  This is a rather
irrelevant matter as far as the file systems are concerned.  There
should be no difference between one computer that owns two disks
and two computers, each of which owns one disk.  The Unix model
works just fine in either case.

As for exec problems, you can easily get them in an isolated Unix
system.  Just install a cross-compiler.  For instance, port the
VAX C compiler to your Sun, and see how well the Sun execs the
output.  Lots of people do things like this.  The distributed case 
adds no extra complexity.

-- 
John Chambers <{adelie,ima,maynard,mit-eddie}!minya!{jc,root}> (617/484-6393)