What to expect: common keyword parameter of data declaration and keyword parameter of constructor have the same name

[PaulKlint]

Setting the stage

Module vis::Graphs contains a data declaration that goes like this (heavily abbreviated):

data CytoLayout(CytoLayoutName name = dagre(), bool animate=false)
    = cytolayout()
    | breadthfirstLayout(CytoLayoutName name = CytoLayoutName::breadthfirst(),   ... )
    | gridLayout(CytoLayoutName name = CytoLayoutName::grid(),  ...)
    | ...;

The striking thing is that name occurs here in two roles:

• as common keyword parameter of the data type CytoLayout
• as keyword parameter of the constructors breadthfirstLayout and gridLayout (and several others)

What does this mean?

Let's morph the above into something really simple

data D(int name = 10)
    = d1(int name = 1)
    | d2(int name = name);

and play with it:

Example	Result	Comment
d1().name	1	ok
d1(name=100).name	100	ok
d2().name	10	ok
d2(name=200).name	200	ok

How does this work on arbitrary D values? Define the getter int get(D d) = d.name; and see what happens:

Example	Result	Comment
get(d1())	10	differs from above
get(d1(name=100))	100	ok
get(d2())	10	ok
get(d2(name=200))	200	ok

It is the different behavior of d1().name and get(d1()) that worries me. It could point at a bug in the interpreter. I am left with the following questions:

1. Is this dual role of name in vis::Graphs accidental or intended. If the latter, what is the intention?
2. Should the type checker detect this situation and warn about it? (At this very moment, letting this pass the Rascal compiler will generate incorrect Java code).