Re: [xsl] Option mystery

["Oleg Konovalov" <olegkon@xxxxxxxxx>]

Guys,

I understood a part of the problem,
that choose is in the for-each block:
<xsl:for-each select="$tier/option">
 <xsl:choose>
   <xsl:when test="$tier/option[condition1]">   <----- bug: I removed
that "$tier/option[]"
   ...
   </xsl:when>
   <xsl:when test="$alternate-tier/option[condition1]">
   ...
   </xsl:when>
 </xsl:choose>
</xsl:for-each>
As soon as I removed that "$tier/option[]", it started working.
The funny thing is that it was in production for 2 years - hidden bug...

But is it technically legitimate to have "$alternate-tier/option[]" in
the second xsl:when block ?
Looks like if similar node exists in parallel tree and matches same
conditions, it sets it's value to "N/A".
Hmm, I am not sure. What do you think ?

Thank you,
Oleg.

On 12/13/06, Michael Kay <mike@xxxxxxxxxxxx> wrote:
> >
> > <xsl:choose>
> >   <xsl:when test="$tier/option[option-num=$cur-option-num and
> > @rate-increase='true']">
> >     <option rate-increase="{@rate-increase}">
> >        <debug1>branch1</debug1>
> >         ...other elements
> >      </option>
> >    </xsl:when>
> > </xsl:choose>
> >
> > A few questions regarding snippet:
> > 1) is that "option" a predefined XSLT1.0 construct / keyword
> > / operation ?
>
> If you mean the option written as "<option rate-increase..." then it is a
> literal result element: an element to be copied to the result.
>
> > I can't find any in the XSLT book.
>
> Look in the index under "literal result element".
> >
> > 2) the bug I am trying to fix is that somehow on exit I am
> > getting a node with <option rate-increase="false"> (that
> > debug1=branch1 and
> > option-num=1).  How is that possible ?   That "branch1" is unique and
> > can't come from any other place. How can that expression
> > inside [] fail ?
>
> Perhaps you're confused because the predicate can only return true if
> @rate-increase is "true", whereas the value that's output is "false". But
> the two occurrences of @rate-option are evaluated with different context
> nodes, so you're looking at two different attributes.
>
>
> Michael Kay
> http://www.saxonica.com/