Robert Dodier wrote:
> ...
>
> Why invent this special convention at all? Why go to the
> trouble of it, when we can just as easily have a symbol
> or expression which explicitly indicates an error?
>
If the error looks like a return value, say "error43" then maxima will
do arithmetic with it, including
error43-error43 --> 0.
>
>> By the way: I am thinking about a way to support the user with
>> additional information about the calculation. One way could be to write
>> messages in a global list, which can be inspected by the user.
>>
This would not be a bad idea, in my opinion, as an additional
technique: e.g.
save all error messages optionally in a global list during a
computation. Then at the end, the list could
be inspected, even if all the messages were squelched during the
computation. Then the list
could be reset. E.g. set error messages "off" but save them on a list
would be an option.
Mathematica sort of has this when it says something like
error xxx
error xxx
... there were 43 more errors error xxx during this computation.
>
> I'm opposed to this. If there is some information which is known,
> just smash it into the return value somehow. If it is too verbose,
> attach it to the property list.
>
It would also be good to have a comprehensive list of all error messages
in the documentation.
RJF