[977] in Athena Bugs

home help back first fref pref prev next nref lref last post

Re: attach: explicit RVD mounts, man-page

daemon@ATHENA.MIT.EDU (Robert S. French)
Mon Sep 19 11:17:54 1988

Date: Mon, 19 Sep 88 11:16:53 EDT
From: "Robert S. French" <rfrench@ATHENA.MIT.EDU>
To: Richard Basch <probe@ATHENA.MIT.EDU>
Cc: bugs@ATHENA.MIT.EDU, bug-attach@ATHENA.MIT.EDU, rt-testers@ATHENA.MIT.EDU,
In-Reply-To: <probe@ATHENA.MIT.EDU>'s message of Mon, 19 Sep 88 11:04:18 EDT
   From: <probe@ATHENA.MIT.EDU>
   Date: Mon, 19 Sep 88 11:04:18 EDT
   Reply-To: Richard Basch   <probe@ATHENA.MIT.EDU>


   This bug applies to 6.0C & 6.0R:

   42% attach -x -e odie:rtinstall
   odie:rtinstall: Illegal explicit definition for type NFS

   First, according to the man page:

	-noexplicit or -x
	     Tell attach to interpret the filesystem as a Hesiod
	     name.  This if the default.

   I thought we were going to keep backwards-compatibility with the old -x
   option for exclusive mode spinups.  Actually, it worked as an exclusive
   mode spinup, so the man-page is in error.

No, we were going to keep backwards compatibility in the Hesiod
information, not the command line options.  The man-page is not in
error; the default for an explicit attach is writable.  The -e simply
cancelled out the -x.

   Second, given the syntax that I used, is it not obvious that I am not
   attempting to access a directory (doesn't a full path have to be
   supplied)?  Thus can't you assume I am attempting to mount an RVD pack
   rather than an NFS filesystem?

When systems software begins to DWIM (do what I mean) instead of DWIS
(do what I said), we are in serious trouble.  When you forget to
supply an argument, or forget a "/", the _worst_ thing that could
happen is the program tries to guess what you meant and then does what
it thinks is right.  The proper behavior is to fail so that you state
your request explicitly (no pun intended).

			Rob

home help back first fref pref prev next nref lref last post