On 13/11/2010 10:29, Thufir Hawat wrote:
> When parsed, why is whitespace counted as a node? It might just be my
> Java bias, but I would greatly prefer the idea of a Tree mapping 1-1 to
> XML. However, with whitespace counting, then the resulting XML text file
> all mushed together.
>
> At least that's my understanding. When parsing XML the parser must
> determine whether or not interpret whitespace as a node or not,
No it must always report it. (Applications may choose to ignore it (cf
xsl:strip-space)
> so that
> different parsers will give different results.
No the behaviour is specified (msxml rather famously ignores the
specification ignoring white space by default, causing all sorts of
problems with inter word spaces vanishing)
Shouldn't the goal be for
> their to be a single correct representation for a given document?
yes
> So
> long as whitespace counts, then it depends upon the parser to either
> include or not include the whitespace.
>
> Is the choice to have whitespace count perceived of as design flaw, or
> does it server a purpose?
If you have <b>this</b> <i>or</i> <b>this</b> and you throw away white
space text nodes then you throw away inter word spaces (as can easily be
seen if you view xml in internet explorer). Not all XML is a database dump
<row>
<item>a</item>
<item>b</item>
</row>
where all white space is cosmetic indentation.
>
>
> -Thufir
>
>
David
