Conference 7.286::digital

    
        Make things easier for the user, not for the
    	computer?
    
    	Where would unix fit in?
    	/john
    	

    ... that's where ...

    Sorry, but .0 is semantically null, imho. What is it trying to say?

    .3
    
    why you say this?
    
     i dont think it sematically null, i think it full of beautifull semantics, 
    all it mean is we the pepole had had it being second to computers, and
    we wont take it anymore, and we will not programme for the computers
    but we will programme for we the pepoles because we were seconds to
    them in the past but in the future we will not be seconds to them
    and we will utilize the means of the computers to improve our hopes
    and dreams and use them to better utilize the benifits of humans 
    beings that lies dorments withen all of us.
    
    i think it all just makes sense to me.
    
    thank you very much,
    /nasser

If writing one's notes entirely in upper case is considered to be
the equivalent of shouting, then surely writing one's notes entirely 
in lower case must be considered to be the equivalent of whispering.

n.

    re: .4 Presumably I can quote you on that for your next review?
    
    	- andy

    re: .3
    
>    Sorry, but .0 is semantically null, imho. What is it trying to say?
    
    .0 is about the following:
    
    The other day my system manager told me one of my conferences'
    (HOME_WORK) disk had a pile of hard errors on it.  It takes up 98% of
    an RD53, so I thought I was in trouble and was going to lose data.
    
    I did an ANAL/RMS on it.  After quite some time (> 1 hour?) it came
    back and told me that the analysis discovered no errors.  Cool.  The
    bad blocks apparently weren't anywhere in the file.
    
    I had an old .COM file around someone gave me that I've been modifying
    for years that does some things I don't understand very well (ANAL/FDL,
    EDIT/FDL, CONVERT/FDL) which compresses notesfiles.  Fine. I run the
    .COM file and it blows up several times because I had set default to
    the wrong disk and it didn't have enough space for work files or
    something.  Whatever.  About six hours later it finishes.  Since
    there's a properly sized new file sitting on my other disk, I delete the
    original.
    
    When I check the new file with NOTES, it barfs on me.  Eh?  I find the
    .LOG for the batch run and it turns out that there WAS a bad block in
    the original file.  The new file LOOKS fine but is actually short.
    
    Since I didn't do any activity on the original disk at all after I
    wasted the file, I look for an UNDELETE command.  There isn't one in
    VMS.  Figuring that there was some trick I didn't know about to getting
    a file back immediately after deleting it, I checked out VMSNOTES,
    VMSNOTES_8, and VMSNOTES_V7.  This takes me back to 30-DEC-1988 in the
    VMS notesfile.  There isn't a single mention of "undelete" in any title
    and searching for "delete" didn't yield anything of interest either.
    
    OK.  No undelete command.  (After all, this isn't MS-DOS V5.0, I
    guess.)  I got my system manager to find me a BACKUP from 6 days
    earlier.
    
    Now I want to make the disk recognize that there are some new bad
    blocks on it.  So I go wandering off into HELP land and find the
    ANAL/MEDIA/EXER COMMAND.  looks like just the thing I need.  When I
    type in the example exactly, it tells me that it can't do what I want
    because  of "revector caching".  After trying to figure out if I can
    turn off such caching using some kind of SET or INIT command, I come in
    and look up the error message.  Here's what I find:
    
    	DEVRCT, unable to analyze 'device-name', due to revector caching
    
    	FACILITY: BAD, Bad Block Locator Utility
    
    	EXPLANATION: The device controller is capable of revectoring 
    	bad blocks dynamically; hence the device will always appear to be 
    	flawless.
    
    	USER ACTION: Do not analyze the media.
    
    FLAWLESS?!?!  I have 238 errors reported on the damn disk!  My
    CONVERT/FDL died because of forced error flags in the file!
    
    So I give up doing this on-line.  I down the machine and boot up my
    "MVII DIAG CUST TK50 COPYRIGHT 1987 DIGITAL EQUIPMENT CORPORATION".
    Half an hour later, the system silently waits for me to enter the date,
    15 minutes after that, it's ready.  I wander through the menus (one of
    which tells me I have to buy some other tape) until I find the disk
    initialization one.  I take it and a while later, it tells me
    (without beeping or anything, of course) that it's done.
    
    Now it didn't tell me that it found any bad blocks or ask me to show
    the bad block list or anything like that and I already KNOW that
    ANAL/MEDIA isn't going to tell me anything so even now, I'm not sure it
    it did the right thing by me.
    
    Anyway, I get the system up, initialize the disk move stuff around and
    everything is finally OK as far as I can tell.
    
    I won't even go into the 2 hour saga of what it takes to figure out how
    to set up a rightslist entry for NOTES$SERVER so you can set an ACL on
    a conference.
    
    Having to go through this scenario in this day and age is what .0 is
    about.  When someone figures out how to make the system call you up on
    the phone and tell you that it debited your account $500 for a new
    disk, installed it itself, and fixed your data while you were sleeping
    (unfiortunately losing note 1435.17 in the process), there will be one
    rich someone out there.  Until then, we're just propagating hacks. 
    Real fancy hacks, but hacks nonetheless.
    
    -joe tomkowitz

    good morning.
    
    ref .6
    >Presumably I can quote you on that for your next review?
    
    by all means , please do , feel free to spread my words.
    
    humm, by the way what review is that? 
    
    buy,
    /nasser
    

    
    
    	Re .7
    
    	Ah ! Welcome to VMS System Management....
    
    	I don't know for sure but I bet there's a 3rd prty vendor
    	out there doing just the things you wanted done....
    
    	Question is why don't we offer a nice system management
    	package ? I admit things have gotten slightly better since
    	V4.0 but not much....
    
    					Former Sysman

    re: .7 Joe, at least you had some meat on the bones. I'm not sure that
    .0 HAS bones.
    
    I agree that intelligent machines are needed to make up for the
    "inadequacies" of users
    
    
    	- andy

    how about an AI based  CLI  too?
    like if a user types $pritn  and they mean $print the computer
    will figure it out and print for them?
    for some reasons someone peoples seem to make a lot of mistakes typing
    and such a CLI will be really neat and help users too.
    
    thank you,
    /nasser

    
    Or a system that will support assists on partial typein, such as
    
    $ PRI?
      PRINT
    
    (Can you say TOP-10? I knew you could!)
    
    (And where have I seen UNDELETE before? Hmmm...)
    

    
 >       (Can you say TOP-10? I knew you could!)
 
	Presumably you intended this as an example of
	"partial type-in" ?

.7 - There are 2 undelete programs available in the TOOLSHED.  They are 
called UNDEL (Note 1473) and UNDELETE (Note 1455).

.12 - Can you say HELP?  I knew you could!

Bruce

    RE .12
    
    Actually, PRI works just fine for PRINT. 

    SO you just drag the document over to the printer icon and drop it.

    
    	If the thing can understand .11 then should be ok, except
    	it will probably crash out on real English.
    
    	:-)
    	

re .11

    how about an AI based  CLI  too?
    like if a user types $pritn  and they mean $print the computer
    will figure it out and print for them?
    for some reasons someone peoples seem to make a lot of mistakes typing
    and such a CLI will be really neat and help users too.
    
    thank you,
    /nasser

There's a public domain UNIX shell named tcsh that does this:

gotit:~> pritn foo
CORRECT>print foo (y|n|e)?

/harley

    ref .18
    thank you harley. that is precisly what i ment by an AI based CLI.
    i agree too with previouse callers that users interfaces is very
    important for an OS to be a sucess in this days of age. 
    i heared it said that NT and win32 is designed to have user in the
    minds and user interfaces. 
    
    goodbuy,
    /nasser

>================================================================================
>Note 1799.14               Power to the users at DEC!                   14 of 18
>AKOCOA::HADDAD                                        6 lines  11-MAR-1992 13:42
>--------------------------------------------------------------------------------
>            -< JOET - A resource you might not have known about.. >-
>
>.7 - There are 2 undelete programs available in the TOOLSHED.  They are 
>called UNDEL (Note 1473) and UNDELETE (Note 1455).
If they are in the toolshed, they are not on my system, so they don't do mw any
good.  They don't get installed by default, so I have to do special things to
get them.  Poor "dumb" user me can't execute the majick to get these
automagically on my system.  That, to me, makes them non-existent.

>
>.12 - Can you say HELP?  I knew you could!
Any "HELP" that requires a special command and clears my screen in the process
of "helping" me hurts more than it helps.  Try some real context-sensitive help
such as is available in Microsoft Windows, TOPS-20, and even EMACS on the
Robin, and you get real help.

>
>Bruce
>
BobW

                                                         
    	Re: .11
    
    	Gee, I know of at least one Digital product that supports exactly
    that kind of command line error correction -- VAXsim V1.0 (and, I
    think, or rather hope, that it still has it).  I wrote its UI out of a
    frusteration that a perfectly reasonable line (continuing the theme
    of picking on VMS ;-) like:
    
    	$ submit/queue=sys$batch/notify/log=[yankes.xyz.logs]/aftr=17:00
    
    would fail since the system was too finicky to figure that "aftr"
    really meant "after" given all the rest of the context of this command
    line.  You know what that user is saying, I know what that user is
    saying, why can't the computer figure it out?  The philosophy of the
    VAXsim parser was to bend over backwards in determining what the user
    really meant to say and, except for the cases where a garbage line was
    converted into something dangerous, just do it.  In the "dangerous" cases,
    the user would be prompted with something like "Is this what you meant?"
    followed by the interpreted line.  (Sorry, I forget the exact comment...)
    Oh, and I put the corrected command line back into the past command
    recall buffer...
    
    	After I was done with VAXsim, I rewrote it in Bliss for generality
    and put it into the toolshed.  Its called the DWIM parser for, yes, "Do
    What I Mean".  (Now if only I can find the sources again.  I probably
    should rewrite this in portable C some weekend... ;-)
    
    	Oh yeah, this was done back in, lets see, 1983?  1984?  This isn't
    new technology that needs fancy AI -- its an easy thing that just
    needs acceptance.
    
    							-craig

    re: .20 
    
    Right on!  If it's in the toolshed, it ain't (by executive fiat, no
    less) in the hands of all of our users.
    
    And right on again!  HELP (with a few minor exceptions or without a
    great deal of prior experience) is worthless.  In my quest for figuring
    out how to find bad blocks on a disk, I eventually figured
    out/remembered that it had something to do with 'BAD' and that it was
    probably going to be a "qualifier" instead of a command so I wound up
    doing:
    
    	$	HELP * /BAD
    
    and then:
    
    	$	HELP * * /BAD
    
    This is a syntax I accidentally stumbled across after years and years
    of trying to get on-line info.  I bet there aren't more than a couple
    dozen people in the world who know that that kind of wildcarding works. 
    As I've said before, with HELP you have to already know what you want
    to ask about it.
    
    re: .21 and the DWIM thing
    
>    This isn't new technology that needs fancy AI -- its an easy thing that
>    just needs acceptance.
    
    I believe that you mean 'user' acceptance.  I would say that it needs
    'corporate' acceptance.  We have these great ideas and even working
    code to do neat things and we still pump out old stodgy boring user
    interfaces.
    
    Is there some kind of corporate edict that says our sexy stuff has to
    be multi-million dollar efforts?
    
    -joe tomkowitz

   RE: .21

   I remember seeing a project/product funding request back in '82 or '83
   for someone to develop an AI-based CLI tentatively title "DWIM".  I 
   thought then how appropriate that would be and how it could be integrated
   into the DCL CLI to help alleviate some (of my!) fumble fingered attempts
   at entering DCL commands.  As time passed, I would occasionally pause and
   wonder what ever happened to DWIM.  It's good to know you succeeded in
   (a) finishing DWIM and (b) incorporating it in VAXsim - Congratulations!

   Now, as .22 mentioned, what can we do to help get "corporate" acceptance
   on DWIM and get it into widespread use?

	Ron

                             
    	Re: .23
    
    	Yup, that was me.  The Software Tools Review Board funded me to
    generalize it and get it into the toolshed.  As to getting it more
    accepted, well, I'm working on another user interface tool (both
    command line and windowing) that will hopefully (eventually) include
    DWIM features when the user is in command line mode.  But, let me put
    this into perspective: this is a wish-list item for something we haven't 
    yet decided to generalize...  Time will tell.
    
    								-craig

    Tops-20 used to do a fair job of dwim-ing... (or at least keeping you
    on the straight and narrow.)
    
    q