[
Lists Home |
Date Index |
Thread Index
]
Hi Evan,
>> All XSLT processors need to govern the building and transforming
>> aspects; xsl:strip-space and xsl:preserve-space give control over the
>> building.
>
> I think you've got the right idea, but a simpler way of looking at
> this is that XSLT processors only have control over the
> transformation, and whitespace stripping is the optional first step
> in any transformation. This view is also more faithful to the 1.0
> spec's description[1].
Well, I guess I think conceiving of it as a four-stage process
(parsing, building, transforming, serializing) is simpler than
conceiving of it as a five-stage process (parsing, building,
whitespace-stripping, transforming, serializing).
I thought that XSLT 2.0 would be describing xsl:strip-space and
xsl:preserve-space in terms of the effect it had on the building of
the node tree. The Data Model WD states that XSLT's whitespace
handling mechanism needs to be supported somehow, so I thought that
would be where it was supported, probably by allowing text nodes to be
designated as insignificant whitespace if they are within an element
whose name is on one list and not on the other, and always setting the
ignore-whitespace flag to true. But obviously you know more than I do
about how the WGs are thinking.
Or maybe I should be picturing it as:
document ---> node tree ---> node tree ---> node tree ---> document
(parsing) (stripping) (transforming) (serialising)
+-----------------------------+
XSLT processor
+----------------------------------------------------------+
XSLT application
And not talking about the PSVI at all; the node tree could be
constructed from some other infoset, after all...
Cheers,
Jeni
---
Jeni Tennison
http://www.jenitennison.com/
|
