Re: [xsl] ancestor

["M. David Peterson" <m.david@xxxxxxxxxx>]

You know there are a few people on this list whom, no matter what my schedule is 
like, I always find the time to read there posts because they always contain 
that extra bit of information or a "clear as a bell" explanation that I had 
never seen or thought of before.  Jeni is one of those people.  It never ceases 
to amaze me when how she can take a small statement and piece of data from a 
question (that to me is completely meaningless) and piece together exactly what 
it is they are trying to do, providing a complete explanation, with several 
examples to boot.

I've been meaning to go and pick up your latest book Jeni, and I think today is 
the day I am going to do that...  I can only imagine how much more depth and 
understanding is contained within its pages.

Thanks for all these years of free education!

Cheers,

<M:D/>

Jeni Tennison wrote:
> Hi Peiyun,
>
>
> > I'm trying to get all inlineequation elements that are not in the table
> > element. The following is not working. I don't know why. I always get all
> > the inlineequation elements.
> >
> > select="//inlineequation[not (ancestor::* = 'table')]"
>
>
> This selects all <inlineequation> elements that do not have an
> ancestor element whose *value* is 'table'. To select all
> <inlineequation> elements that do not have a <table> element as an
> ancestor, you can use:
>
>   select="//inlineequation[not(ancestor::table)]
>
> The path "ancestor::table" selects the <table> ancestor element, if
> there is one.
>
>
> > I know this can be very expensive even if it works. What cab be a
> > better way to do it?
>
>
> The thing that's expensive here is the use of the descendant-or-self::
> axis (implicit in the "//" at the beginning of the path), which means
> that the processor has to search the entire document for the
> <inlineequation> elements. Anything that you can do to narrow down the
> search for the processor will help. It might be that you can search
> only within the element that you're currently on (use
> ".//inlineequation") or that you can use the natural recursion of the
> templates to get these elements.
>
> If you have lots of tables, for example, then rather than doing:
>
>   <xsl:apply-templates select="//inlinequation[not(ancestor::table)]"/>
>
> it might be more efficient to use templates to filter down the
> subtrees that get searched. Use modes -- in 'inlineequation' mode,
> you're only interested in <inlineequation> elements. Apply templates
> to the top of the tree:
>
>   <xsl:apply-templates select="/" mode="inlineequation" />
>
> and then have a template for <inlineequation> elements:
>
> <xsl:template match="inlineequation" mode="inlineequation">
>   ... do something ...
> </xsl:template>
>
> Let the built-in templates do their work (recurse down the tree), but
> have templates to stop processing whenever you come to an element that
> can't contain an <inlineequation> element that you're interested in,
> such as a table:
>
> <xsl:template match="table" mode="inlineequation" />
>
> This assumes that you really want to collect together all the
> <inlineequation> elements -- that you're creating a "List of
> Equations" or something -- if that's not what you're doing, then
> there's probably a more efficient approach, but it would require
> looking at more of your stylesheet to know what's suitable.
>
> Cheers,
>
> Jeni
>
> ---
> Jeni Tennison
> http://www.jenitennison.com/