[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Logical names

    Date: Saturday, 27 April 1985, 20:12-EST
    From: Robert L. Krawitz <RLK%MIT-OZ@MIT-MC.ARPA>

    In ZWEI in System 98.80, CADR 3.10, ZMail 53.19, MIT-Specific 22.5,
    microcode 309, ZM MIT gc*2, on Ford Prefect:

    This keeps happening to me.  It's a problem with logical devices on

    The default file name in zwei was ee:ps:<le.foo.bar>bletch.baz.  I
    wanted to save something to ee:lpt: (in other words, to the printer).
    It came back with error invalid wildcard designator in
    ee:lpt:<le.foo.bar>bletch.baz.  I then tried ee:lpt:<> and it came back
    with error ee:lpt:<>bletch.baz.  ee:lpt:<>..0 failed on
    ee:lpt:<>*buffer-2*..0.  It took some hair to make it work.

    Sorry, no backtrace -- the window it tried to use to mail that bug
    report was arrested in some hairy fashion and wouldn't unarrest even
    with the mouse, system menu, etc.

    1)  WHY does it try be so fancy with the logical names?  
It doesn't.  It never HEARD of them.
							     It tries to be
    so lazy-mode with defaulting that it refuses to accept the possibility
    that I @i(want) a null directory or filename string.  
Right.  Unless you can arrange for the lisp machine end to find
out from the TOPS-20 somehow which device names don't have directories
with them, there is no way to have both defaulting, which is universally
useful, and being able to access certain special devices, which is
useful on rare occasions.  As for logical names (which your problem
wasn't really with), the same problem exists; knowing when something
is a logical name with a directory supplied, as opposed to the PS: vs
SS: vs SRC: type distinction, which is much more common.

							  The fact that this
    is a special case is no excuse -- it should accept special cases
    gracefully.  It shouldn't default unless the user explicitly hits the
    altmode key.  Automatic defaulting loses when someone doesn't want
    something to default.  Twenex at least handles it right -- if you don't
    type anything it assumes you don't want it.  
This is not how TWENEX handles it.  TWENEX has all the cards; it knows
when something is a device, when it is a logical name, and when it is
just the name of the other structure.  If it is just the name of another
file structure, it does the same defaulting the lisp machine does.

						 If I can set up a backtrace
    I will.
Don't bother.

    2)  I have similar types of lossage with all the logical names I have
    defined (i. e. one for each directory).  So if I ask for a dired for
    bit: which is <zzz.rlk.bitnet> it tries to dired bit:<zzz.rlk> and does
    oz:<zzz.rlk>.  Why can't it assume I mean what I tell it?  

Because you told it to take the default for the directory, and you
never told it BIT: was a logical name.  You told that later essential
bit of information to OZ, not your local machine.
							       Even if the
    spec I give it isn't technically correct for twenex, it shouldn't assume
    that I mean something different from what I tell it.  If I ask for
    completion it can always do it THEN.

This would be a highly inconsistant user-interface.  Not only would it
be inconsistent with how every other kind of pathname works on the lisp
machine, it would even be completely inconsistant with how it works on