At Tue, 21 May 1996 08:10:29 -0700, Bruce Toback <[log in to unmask]> wrote:
[Neat comparisons about Saturn V snipped...]
>Ah, then what you want is a Unix or PC solution. The standards are
>changing, and us dinosaurs (over-25s) had better change with them. Today,
>"good enough" IS good enough. People expect to lose some data now and
>then; hasn't "sorry, my computer's down" become as accepted an excuse as
>"sorry, my car wouldn't start"? Read the newspaper "computer advice"
>columns, or go to an "Internet" class. Having the right answer 90% of the
>time is perfectly acceptable;
[...]
>By and
>large, a crash here and there or a lost file or an inexplicable
>computational result are just part of everyday living, like dead
>batteries.
>
 
Ah yes,  what we called in the old days, "Management By Disaster".   That
with my other favourite practice, "Implement and Run!" made for a lot of fun
back then. :)
 
But seriously, what Bruce says has a lot of merit to it.  In studying User
Interface design, it can be obvious where screens have been designed by
computer techies, rather than users.  For example, consider the case where
there are two possible choices: one used 99.99% of the time, the other only
0.01%.  The techie will clutter the screen with both choices, whereas the
user will see only want the common choice, with the alternative removed to a
drop down menu or dialog somewhere else.  For the techie every choice is
equally valid, because the techie must program it that way.  The user though
when dealing with limited space only wants to see the standard choices, but
have provision for exceptions.
 
There's an old joke:
  Man in a bar to female bar fly, "Would you go to bed with me if I gave you
                                   a million dollars?".
  Woman: "Sure!".
  Man: "Would you go to bed with me if I gave you five dollar?".
  Woman: "No way, what do you think I am?".
  Man: "Well we know that...now we're just haggling on the price.".
 
The point being, even if we spend an almost limitless amount of money, we
can't *guarantee* that an HP3000 won't crash: it could be operating
software, or third party software, a memory failure, or a plane crash into
our system. Even with full Shareplex, raid and other redundancies, there is
still a remote possibility that the system can crash.  The system is
tainted.  So perhaps rather than spending a lot of money on making a system
that is more stable and reliable, we should look at ways at operating with
unstable and unreliable systems (like Unix and NT?).
 
Naturally, for some systems the crash can be more painful than others - the
tradeoff is in the pain it causes, the delay in restarting and the value of
the data lost.  So maybe some systems can't be so lax - but there is plenty
of scope for others - the drift I see is the number of absolutely critical
non-fail systems is becoming less and less, while more alternatives are
presenting themselves for minimizing crash-recovery times.
 
<Total.................. $0.02>
 
----
Jim "seMPEr" Wowchuk
Vanguard Computer Services     Internet:    [log in to unmask]
 _--_|\                        Compu$erve:  100036,106
/      \                       Post:        PO Box 18, North Ryde, NSW 2113
\.--.__/ <---Sydney NSW        Phone:       +61 (2) 888-9688
      v      Australia         Fax:         +61 (2) 888-3056