[bug-libmatheval] Some improvement proposal

[Romain Garbi]

Hi !

I was a little bit long to answer (well, was an answer really needed ? 
not really because yours solved my issue).

Le 23/12/2013 09:11, Aleksandar Samardžić a écrit :
> It's natural logarithm - libmatheval is following C language math
> library naming conventions.

Ok, thanks. Anyway I think it is worth mentioning that in the 
documentation and in constants, natural logarithm is sometimes referred 
to as ln(). Maybe it's a mistake, or maybe not, but that was what 
confused me ^^

Moreover, I'd like to submit an idea : why not include an evaluate 
function that use a callback function to get variable values ?
I think such a function might be probably more convenient for developers 
who want to link a project that already handles variables
internally, because it avoid them to write a subroutine to build the 
arrays needed by evaluator_evaluate.

I've already the code for the function:

double
evaluator_evaluate2(void *evaluator, double(*function)(const char*))
{
        Record         *record; /* Symbol table record corresponding to
                                 * given variable name.  */
        int             i,      /* Loop counter.  */
                    count; /* number of symbols in names */
        char**          names; /* symbols name */

    /* Get expression variables */
        evaluator_get_variables(evaluator, &names, &count);

        /* Assign values to symbol table records corresponding to
         * variable
         * names. */
        for (i = 0; i < count; i++) {
                record =
                    symbol_table_lookup(((Evaluator *) evaluator)->
                                        symbol_table, names[i]);
                if (record && record->type == 'v')
                        record->data.value = function(names[i]);
        }

        /* Return requested value */
        return node_evaluate(((Evaluator *) evaluator)->root);
}

Finally, I was wondering if it was possible to specify a prefix for 
symbols that are produced by Lex and Yacc (using -P and -p switches) to 
avoid unnecessary conflicts between two code generated by Flex and Yacc 
linked together.

Regards
Romain