Re: [xsl] Antwort: RE: [xsl] bad programming for speedup?

[Robert Koberg <rob@xxxxxxxxxx>]

Hi,

Is the OP sure that the elements wrapping the rows cannot be determined
beforehand and have templates set up for those? If so, from what I can
see, all that is needed is:

<xsl:template match="tbl-type1 | tbl-type2 | tbl-type-etc">
  <table>
    <xsl:apply-templates/>
  </table>
</xsl:template>

<xsl:template match="node() | @*">
  <xsl:copy>
    <xsl:appy-templates select="node() | @*" />
  </xsl:copy>
</xsl:template>

best,
-Rob


On Tue, 2007-07-24 at 14:46 +0200, Abel Braaksma wrote:
> Hi Christoph,
> 
> Have you tried the other suggestions? Mainly Andrew's method (or mine 
> even, if the below does what I think it does).
> 
> Please allow me to make some suggestions on your code below:
> 
> 
> 
> christoph.naber@xxxxxxxxxxxxxxxxxxx wrote:
> >   
> >       <xsl:template match="/">
> >                 <xsl:apply-templates />
> >         </xsl:template>
> >   
> >
> 
> This is not necessary in this particular template. However, you may feel 
> that having a specific entry point in your code is a matter of good 
> practice (many do).
> 
> >  
> >         <xsl:template match="*">
> >                 <xsl:copy>
> >                         <xsl:copy-of select="@*" />
> >                         <xsl:apply-templates select="*" /> 
> >                 </xsl:copy>
> >         </xsl:template>
> >   
> 
> To save yourself some keystrokes (also further on), you can as well 
> write the above as follows:
> 
> <xsl:template match="node() | @*">
>     <xsl:copy>
>        <xsl:appy-templates select="node() | @*" />
>     </xsl:copy>
> </xsl:template>
> 
> In addition, this way (using node() instead of *), text nodes, 
> processing instructions etc are also processed by your copy template.
> 
> >  
> >         <xsl:template match="row[position() = 1 or 
> > preceding-sibling::*[name() != 'row' and position() = 1]]" >
> >                 <table>
> >                         <xsl:apply-templates select="." mode="more2come"/>
> >                 </table>
> >         </xsl:template>
> >   
> 
> Take a close look at the way Andrew coded this same principle. Also, 
> consider an earlier remark by Justin about *not* using name() to compare 
> nodes. Use self::row instead. This will help you when you are attempting 
> sources that contain namespaces.
> 
> >         <xsl:template match="row" mode="more2come">
> >                 <xsl:copy>
> >                         <xsl:copy-of select="@*" />
> >   
> 
> The line above (xsl:copy-of) can be removed if you follow the remark 
> made above about the copy template.
> 
> >                         <xsl:apply-templates select="*" /> 
> >   
> 
> And this should really read:
> 
>    <xsl:apply-template select="node() | @*" />
> 
> to make it all work well.
> 
> >                 </xsl:copy>
> >                 <xsl:if test="following-sibling::*[name() = 'row' and 
> > position() = 1]">
> >                         <xsl:apply-templates 
> > select="following-sibling::row[position() = 1]" mode="more2come" />
> >                 </xsl:if>
> >   
> 
> Whenever you find yourself making an xsl:if with a node test, followed 
> by an xsl:apply-templates with (largely) the same call, it is most 
> likely that you can easily write it in one statement and make it clearer 
> for you and fellow coders. Note that xsl:apply-templates does nothing 
> when the selection contains an empty set, making the xsl:if from 
> imperative programming style largely redundant.
> 
> See the code of Andrew for an example.
> 
> 
> >         </xsl:template>
> >  
> >         <xsl:template match="row" >
> >         </xsl:template>
> 
> Using the approach of Andrew, you do not need this delete-template, 
> saving on keystrokes again ;)
> 
> If you want it simple and fast, I believe the approach with keys is 
> quicker, but not necessarily so of course. In addition, since all you do 
> is plain copying and no further processing, it is likely that the 
> <xsl:copy-of> constructor from my example will work the fastest 
> nonetheless. But I haven't seen your source, so I couldn't tell (and 
> with an optimizing processor it may matter zero after all).
> 
> Cheers & happy coding ;)
> 
> -- Abel Braaksma