[Bug 1705] [FS] technical: 5.2 Module Declaration: cyclical imports

http://www.w3.org/Bugs/Public/show_bug.cgi?id=1705





------- Comment #13 from jmdyck@ibiblio.org  2006-10-11 08:28 -------
> About #10. My understanding is that the XQuery 1.0 specification does not
> impose a particular way to implements the circularity detection.

Supposedly, it doesn't impose a particular way to implement *anything*, it just
says that an implementation has to get the specified results. I guess what you
mean is that, with respect to circularity detection, it's not even saying what
the specified results are. So perhaps the exact meaning of "circularity" is
effectively implementation-dependent.

> It does not have to be done through the static context as defined in the
> XQuery Formal Semantics, although a clever implementation might try and
> do that. I think what the resolution suggests is that this is not formally
> specified in FS.

That's not surprising, since the FS doesn't formally specify the conditions of
most other static errors either. But this bug isn't really about that.
Evaluation-circularity is only relevant to this bug insofar as it could
conceivably be defined in a way that would outlaw certain cases that otherwise
might be hard to do STA on.

Specifically, if my example were deemed illegal (due to circularity, or
whatever), then the FS certainly doesn't need to give a STA for it. But if, as
you suggest in Comment #11, it *is* legal, then presumably the FS is obliged to
say how one could do STA on it. The current FS doesn't. Neither does my
original suggestion. So if the next version of the FS merely adopts my
suggestion, then it probably won't either.

Received on Wednesday, 11 October 2006 08:28:30 UTC