[alexeym]

During ongoing attempt to build OpenFOAM with "-Wall -Wextra -std=c++11" flags without any warnings the following questions arose. Since they are not bugs and not a feature proposal yet, I have decided to discuss it here.

1. C++11 standard deprecates register variable modifier. Yet it quite frequently used in base library. Should it be kept in the code?

2. There are plenty of unused parameters in the code, which way of suppressing warning is preferred? There are several common methods: use UNUSED macro (which is basically UNUSED(x) (void)x), comment-out name of the parameter, just remove parameter name.

3. Though both g++ and clang produce code that can deal with NULL-pointer dereferencing into constant reference (directAddressing in FieldMapper.H is an example), stiil this behavior is undefined in standard. Should these parts kept as they are or should pointer-returning methods added to those cases?

4. In most cases implementation of child classes is OK, but from time to time developers desire strange things (functionObjects library is an example) like return type overload. I.e. functionObject class is a child of regIOobject, yet it is not happy with write method that returns bool, it would like to have write method returning nothing (i.e. void). Should such cases be prohibited (i.e. enforce correct method overload) or left as they are (allow method hiding)?

5. This question is just about author intentions in readKivaGrid.H file. There is a loop on line 426

        for
        (
            ;
            iterf != pFaces[WEDGE][0].end(), iterb != pFaces[WEDGE][1].end();
            ++iterf, ++iterb
        )
        {
	...
	}

comma operator discards left expression, i.e. result of the first comparison is just ignored. Should it be

        for
        (
            ;
            iterf != pFaces[WEDGE][0].end() && iterb != pFaces[WEDGE][1].end();
            ++iterf, ++iterb
        )
        {
	...
	}

?

6. This question is about author intention in surfaceBooleanFeatures.C. On line 357 there is calcFeaturePoints function. At the end of this function the following variables are defined:

    label concaveStart = 0;
    label mixedStart = 0;
    label nonFeatureStart = nFeaturePoints;


    labelListList featurePointNormals(nFeaturePoints);
    labelListList featurePointEdges(nFeaturePoints);
    labelList regionEdges;

and after these definitions function just ends. Is this just unfinished function or these definitions are there for particular reason?