MathGroup Archive 1998

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Bad error messages

  • To: mathgroup at smc.vnet.net
  • Subject: [mg14132] Re: [mg14087] Bad error messages
  • From: David Withoff <withoff>
  • Date: Mon, 28 Sep 1998 03:51:47 -0400
  • Sender: owner-wri-mathgroup at wolfram.com

> At Warwick University we have a site license for Mathematica which runs
> off a license server. I don't know the details of how this works or
> doesn't work, but the license server does seem to collapse quite
> frequently. If I am using Mathematica when the license server goes
> down, I get truly bizarre error messages. This was at first totally
> puzzling and mystifying, particularly as the next time I ran the same
> program everything would work perfectly.
>
> When I wrote to support at wri.com about this (before I diagnosed the
> reason for the difficulty), I got no response except an automated
> acknowledgement that my email had been received.

Something must have gone wrong with this communication.  It would be
good to know what it was.  We have no record of messages from
dbae at maths.warwick.ac.uk, and a recent message (September 14) from
another address at warwick.ac.uk, about a license server problem, was
answered that same day.

If you don't get a reply from support at wolfram.com in a reasonable amount
of time you should suspect a communication problem.  There are lots of
things that can go wrong with email communication.  We recently tracked
down a problem, for example, in which anti-spam software at the users
site was disarding our replies.  I'm not suggesting that that is what
has happened here, but it is an example of one of the many reasons that
a message may have gotten lost.  If you suspect that a message has
gotten lost, check the email address and try sending another message,
or try communicating in some other way, such as through another
address, by telephone, or by whatever means is appropriate for your
situation.  The correct address for Wolfram Research Technical Support
is support at wolfram.com, but the address that you mentioned
(support at wri.com) should still work.

> It seems to me that WRI should revamp the error message system. If your
> license is not valid, shouldn't the error message say this rather than
> come out with arbitrary reports of failures of arbitrary aspects of the
> system.

As someone who has spent considerable time both answering questions
about error messages and working to improve error messages I have a
special perspective on both the importance and the difficulty of
addressing this concern.  The core of the difficulty is that, when
something goes wrong, the computer never knows what the problem is; it
only knows the symptoms. Sometimes a message can offer speculations
about the problem, as in

In[1]:= NIntegrate[Sin[x],{x,0,2Pi}]


NIntegrate::ploss:
   Numerical integration stopping due to loss of precision. Achieved
neither
    the requested PrecisionGoal nor AccuracyGoal; suspect highly
oscillatory
    integrand, or the true value of the integral is 0. If your integrand
is
    oscillatory try using the option Method->Oscillatory in NIntegrate.

                   -17
Out[1]= -5.55112 10

and sometimes the computer can do further analysis automatically.  Such
speculation and analysis can be misleading if not done carefully, and
making it work well is an endless task.  Of special concern for license
server messages is the fact that the number of things that can go wrong
is limited only by the number of ways in which a computer system can be
configured.

Despite these difficulties, you certainly won't get any disagreement
about the desirability of generating useful error messages whenever
there is a realistic way to do it.

> David Epstein
> University of Warwick

Dave Withoff
Wolfram Research


  • Prev by Date: Copy and Paste
  • Next by Date: Conference Videos Now Available
  • Previous by thread: Bad error messages
  • Next by thread: Find combinations in k-subsets