Re: [xsl] bad programming for speedup?

[Abel Braaksma <abel.online@xxxxxxxxx>]

Hi Christoph,

I very much wonder what processor you are using, as the "bad" version 
(and it is indeed _bad_ ;)   Can you tell a little about what 
environment you use: XSLT version (probably 1.0, see list guidelines if 
you are not sure how to find out), what processor and what processor 
version? Do you use it in a browser, if so which? Do you call it with PI 
or from a commandline or from something else?

see also my comments below,

Cheers,
Abel Braaksma

christoph.naber@xxxxxxxxxxxxxxxxxxx wrote:
> Hello everybody,
>
> I'm new to this list, so let me introduce myself: 
> My name is Christoph Naber. I come from Germany and work with XML/XSL for 
> about 2 years now.
>   

Welcome!

> I'd be pleased to hear your opinion about a stylesheet I've written.
>   

fasten your seatbelts ;)

> The aim is to surround occurences of <row> - tags with a <table> - tag. 
>   

opinion 1: XSLT does not know anything about "tags", it works with nodes 
instead. After serialization, this may or may not translate into opening 
and closing tags.

> I've done this with "good" XSL, what appears to be real slow
>   

I'd love to see your "good" XSLT, I have yet to see something that is 
indeed slow

> , and with a 
> "bad" version, which inserts tags as <xsl:text>.
>   

I completely concur with your idea of "bad". I would even say: "evil", 
as it is very dangerous to use d-o-e (short for disable output escaping) 
and to rely on it creating your < and > to create tags. You should let 
the serializer take care of that (which is why I am interested in your 
good version). The one reason I can think of why your "good" version 
might be slow is that it builds a very big internal tree (i.e., larger 
than the available memory), otherwise, the speed should not really be 
different (but then again, I have no idea what processor you are using 
so I cannot really tell).

> The second version:
> <xsl:template match="row" >
>         <xsl:if test="name(preceding-sibling::*[1]) != 'row'">
>                 <xsl:text 
> disable-output-escaping="yes"><![CDATA[<table>]]></xsl:text>
>         </xsl:if>
>         <xsl:text>
> </xsl:text>
>   

You are testing for the first "row" to be there and you are writing your 
own tag??? I am very surprised it works even. D-O-E is not guaranteed to 
be supported, and even if it is supported, it is not guaranteed to 
create the output you seem to want.

>         <xsl:copy>
>                 <xsl:copy-of select="@*" />
>                 <xsl:apply-templates select="*" /> 
>         </xsl:copy>
>  
>         <xsl:if test="name(following-sibling::*[1]) != 'row'">
>                 <xsl:text 
> disable-output-escaping="yes"><![CDATA[</table>]]></xsl:text>
>         </xsl:if>
>   

ouch, this hurts my eyes ;) . Same here.... as above....


Again, I wonder why it works and even more, I wonder what your good code 
looks like. Here's a version that should work (many variants possible, 
this one is *very fast* if you do not need extra processing of children 
and it is supposed to be faster than your _bad_ version):

   <xsl:template match="/">
       <xsl:apply-templates select="your/path/to/row[1]" />
   </xsl:template>

   <xsl:template match="row">
       <table>
           <xsl:copy-of select=". | following-sibling::row" />
       </table>
   </xsl:template>


Cheers,

-- Abel Braaksma