Re: [WebIDL] cycles in [PutForward] chains

On Sat, Aug 4, 2012 at 11:33 AM, Tab Atkins Jr. <jackalmage@gmail.com>wrote:

> On Sat, Aug 4, 2012 at 9:50 AM, Glenn Adams <glenn@skynav.com> wrote:
> > In [1], the following language appears:
> >
> >> Note that [PutForwards]-annotated attributes can be chained. That is, an
> >> attribute with the [PutForwards] extended attribute can refer to an
> >> attribute that itself has that extended attribute. Theremust not exist a
> >> cycle in a chain of forwarded assignments. A cycle exists if, when
> following
> >> the chain of forwarded assignments, a particular attribute on an
> interface
> >> is encountered more than once.
> >
> >
> > While this states that a cycle must not exist, it fails to define
> > implementation behavior regarding cycle detection and action in the
> presence
> > of a cycle. If this is intentionally left undefined, perhaps that should
> be
> > stated.
> >
> > [1] http://www.w3.org/TR/WebIDL/#PutForwards
>
> Since the audience for that requirement is spec authors themselves...
>

I think it is not merely spec authors but implementations, since the
language above says "when following the chain". For example step 6 of [2]
constitutes following the chain, which is a runtime semantic.

[2] http://www.w3.org/TR/WebIDL/#dfn-attribute-setter