[comp.os.vms] What's a good policy for system-wide logical names?

IJAH400@INDYVAX.BITNET (05/27/88)

I never saw this posted, so I am trying again.  Sorry if you have seen it
before...

Hello out there all you VMS system managers!  We are completing a conversion
from another Digital architecture and operating system, and have run into a
little problem concerning logical names.  It is a swell (well, in some cases
I suppose) feature of RMS that one of the first things it tries to do with
a filespec containing no punctuation is to iteratively translate it as a
logical name.  Now, once in a while a user will get confused by this because
they will have some logical name X defined, perhaps pointing to a directory,
or some other unrelated file, and then try to use the name X in some other
context.  Perhaps they have a FORTRAN program named X.FOR, and try to compile
it with the command:

$ FORTRAN X

If X is, say pointing to an object library or something else wierd, the
FORTRAN compiler will not like it at all.  If the user is responsible for
the definition of X, then he obviously deserves what he gets.  But, if he
isn't, and if he (or she) isn't aware of the logical name definition, it can
be quite confusing.

Now, Digital has reserved "$" for use in their logical names exclusively.
As we have installed third party software, we have noticed that some vendors
seem to try to keep some sort of pattern, like starting all their logical
names with a prefix and an underscore (some even use the sacred dollar sign).
Others (MACSYMA is an example, with names like TMP, BIN, etc.) don't seem to
care one whit.  What we have been doing so far for our site-specific system-
wide logical names has been a mixture of both.  The other day we had our
first confused user consulting problem, who unfortunately is a professor
who complains rather loudly.

I am going to pretend I have no preference on this issue.  What I would like
is the sage advice of others who have experienced this problem and how they
dealt with it.  Is the convenience of short names worth the consulting
headaches caused by them?  Are the users who whine about having to type long
names louder than the ones who complain about getting confused by the short
names, or vice-versa?  Do users strive for ever-longer names until they find
one long enough to get confused over anyway? (i.e., is it a problem with
unsophisticated users no matter what policy one adopts?).

In order to keep the number of messages bouncing around on the net down,
please send your replies directly to me, and I will summarize the responses
to the net in two weeks (or sooner if my disk fills up with mail).  Please
indicate the average sophistication of users at your site (i.e., do most
of them even *know* what a logical name is? -- this would appear to affect
the magnitude of the consulting problem) and some indication of site size
(number of users) would be helpful if possible.  My BITNET address is:

        IJAH400@INDYVAX

Many thanks to all who reply in advance.

 - James A. Harvey, Senior Analyst/Programmer, DEC Systems Group
   Computing Services, Indiana University/Purdue University at Indianapolis
   (IUPUI)

   Bitnet: IJAH400@INDYVAX

   Disclaimer: "Any opinions offered here are certainly not those of
                my employer, as we never have agreed about anything!"