[pygccxml-development] FT argument\return type\value indexing - decision

[Roman Y. <rom...@gm...>]

Hi.
I think we discuss enough this point and there is no sense to continue. I am
not
happy with the way I do it, but it seems that I have no other choice.

Problem:  how user should address\refer to function argument and return
type\value.

Current implementation: index is 1 based, in order to allow user to specify
0 as
return.

The solution:

1. Index should be 0 based.

2. Special const will be introduced to represent "reference" to return
type\value:
    return_ = None ( or -1 ).

3. function_transformation_t will have new variable - USE_1_BASED_INDEXING
    By default its value will be False. Every one who consider "more
natural" to use
    0 based indexing will set this to True.

4. function_transformation_t will introduce 2 new methods:

    def get_argument( reference, func ):
        if isinstance( reference, str ):
            found = filter( lambda arg: arg.name == reference,
func.arguments )
            if len( found ) == 1:
                return found[0]
            raise RuntimeError( "Argument with %s was not found" % reference
)
       else:
           assert isinstance( reference, int )
           if function_transformation_t.USE_1_BASED_INDEXING:
               reference += 1
           return func.arguments[ reference ]

    def get_type( reference, func ):
        if isinstance( reference, int ) and reference == return_:
            return func.return_type
        else:
            return get_argument( reference, func ).type

    As you can see new implementation allows to use argument names as
parameter
    to transformers.

5. Every built-in transformer will have to use these functions.

One more time, I prefer to make a decision, when there is a consensus.
This time we don't have it. I hope the solution provides a compromise.
We still may discuss any detail of the solution, but I would like to make it
clear:
by default indexing will be 0 ( zero ) base.

-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/