CoCoA » CoCoALib

Here is note for accumulating ideas for structuring CoCoALib exceptions:
(1) create CannotDivide which has two subclasses: InexactDivision and DivByZeroDivisor
(2) create IndexOutOfRange with subclasses BadRowIndex and BadColIndex

• IndexOutOfRange with suppl mesg "matrix row index"
• UnsuitableValue with suppl mesg "value must be >= 0"
• BadType with suppl mesg "defined only for FractionFields"

Comparison of error objects will completely ignore the suppl mesg; so two IndexOutOfRange error objects with differing suppl mesgs will compare as equal. This is so that one can perform a check like if (ErrObj == IndexOutOfRange) ...; WARNING it may produce some unexpected results if error objects are used in std::map containers.

What is a reasonable syntax for adjoining a suppl mesg to an error object?

(A) Anna suggests a fn-call style: IndexOutOfRange("matrix row")
(B) John suggests a binary op: IndexOutOfRange + "matrix row"
(B*) variant of proposal (B): IndexOutOfRange += "matrix row"
(C) BOOST uses a syntax like that for ostreams: IndexOutOfRange << "matrix row"

What happens if one tries to add several suppl mesgs?
(1) gives an error
(2) suppl mesgs are appended
(3) only the last suppl mesg is recorded
(4) only the first suppl mesg is recorded

JAA does not like option (1): getting an error while trying to prepare an error message is not ideal!! Option (2) is probably simplest to implement (and possibly the most natural) -- do we want to encourage/discourage appending multiple suppl mesgs? JAA thinks that the BOOST syntax encourages it.

NOTE (2013-09-11) if we want option (2) then JAA thinks syntax (B) or (B*) is more natural than syntax (A)

There are some notes about "classes of errors" in the CoCoALib documentation for error, section "=== new improved list of errors ==="

Anna Maria Bigatti wrote:

There are some notes about "classes of errors" in the CoCoALib documentation for error, section "=== new improved list of errors ==="

I suggest keeping the errors in error.C sorted as in that list (instead of the current alphabetical order): I believe that in this way, we would consider the problem of classifying the errors when we are thinking/looking for an error message, so when our brain is more likely to find a good strategy.