[Bug 29983] [XSLT30] Scanning expressions and function calls

https://www.w3.org/Bugs/Public/show_bug.cgi?id=29983

--- Comment #10 from Abel Braaksma <abel.braaksma@xs4all.nl> ---
> Re comment#9:
We used to have the term *equivalent expression* but decided to drop it because
we couldn't define it specifically enough. I believe the problem was that we
define clearly to go from a pattern to an expression, but the inverse is not so
simple and would lead to processor-dependent behavior on what expression also
is a motionless pattern (I think it was in the discussion of bug 25160).

So, if there are no differences, that is good, then the definition is as it
should be. But I'm uncertain we should go back to a situation we dismissed
previously as too vague.

That said, if we can convince ourselves it is not vague and that for all E
(expression) in EE (all RelativePath expressions) we can define unambiguously
whether it can be rewritten as a motionless pattern or not, it would be a good
way forward into simplifying this. But I have a strong feeling that precisely
that discussion led to the introduction of scanning expressions in the first
place.

Note that, with the present rules, an expression like //foo and the pattern
//foo have distinct streamability semantics, the patterns are streamable, and
the expression is only streamable in rare cases (if the context type is
document-node), meaning that the sets of motionless patterns and streamable
RelativePath expressions is not (presently) equal.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.

Received on Thursday, 17 November 2016 23:35:38 UTC