I, for one, find the current messages perfectly adequate but I suppose this
is my fault for not being a clueless newbie.
Compare some equivalents to your examples in Excel:
Faulty Expression Error Message
------------------------ ---------------------
=3+'A' The formula you typed contains an error.
o For information about fixing common
formula
problems, click Help
o To get assistance in entering a
funcion, click
OK, then Function on the Insert menu
o If you are not trying to enter a
formula, avoid
using an equal sign (=) or minus sign
(-), or
precede it with a single quotation
mark (').
=randbetween(6,5) #NAME? [if Analysis Toolpack not installed]
=randbetween(6,5) #NUM! [if Analysis Toolpack is installed: the arguments
are supposed to be bottom, top of range]
=rand(6,5) [Same as first]
=rand('A') [Same as first]
=sumproduct(C2:C3,D2:D5) #VALUE!
In the first case, the initial part of the message is completely vague and
unhelpful; the remainder of the message is generic boilerplate for a wide
range of possible errors having nothing to do with the specific problem.
The other messages that are different at least have the virtue of brevity
though they are similarly unhelpful. The last example, with "sumproduct", I
chose because it takes arrays as arguments. In this case, the message is
unhelpful to the point of being misleading: when I first got it, I suspected
it might have to do with the arrays being empty cells. However, filling in
the cells with values did not change the message, so it really is a "length
error".
While it might be nice to implement the wordy, more specific messages (Ed)
proposed as an adjunct help system to which a beginner could refer, it would
infantilize the rest of us. Again, this looks like the choice between
hindering the expert user to provide a transitory benefit to the novice or
asking that people actually think about what they are doing.
A sharp knife is dangerous to the inexperienced but indispensible to the
expert cook; a dull knife is dangerous to everyone and not a very good tool.
On 12/15/07, bill lam <[EMAIL PROTECTED]> wrote:
>
> In APL89, Edward J Shaw contributed a paper with this title. He argued on
> the
> subject "user-friendly or theoretically correct", examples transliterated
> to J
>
> errant current(in 1998) proposed
> -------------------------------------------------------------
> 3 * 'A' domain error improper value - argument must be numeric
> 6 ? 5 domain error improper value - value of left argument may
> not
> exceed value of the right
> argument
> 3 + (5 syntax error syntax error - unmatched parentheses
> 2 3 + 4 5 6 length error shape error - length of corresponding axes
> must be equal
>
> so far nearly two decades years later there is little change, at least in
> J.
>
> --
> regards,
> bill
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
--
Devon McCormick
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
