substring(lduk_vp_date,1,10) !=
substring(preceding-sibling::*[1]/lduk_vp_date, 1, 10)

However you should note that the preceding-sibiling:: axis applies to the
order of the nodes in the document.  Hence this only works if the all
transactions for a specified day are adjacent to each other.  That being the
case your xsl:sort is probably redundant.


"IanK" <iankiddcroftamie@y...> wrote in message
news:1175764392.551602.275300@y......
> I am trying to produce output like this:
>
> 01/01/2007  Something for january 1st
>             Something else for january 1st
>             And another for january 1st
> 05/01/2007  A data row for january 5th
>             And another for january 5th
>
> And I'm using code like this:
> <xsl:for-each select="transactions/transdet">
>   <xsl:sort select="lduk_vp_date"/>
>     <tr valign="top">
>       <td align="left">
>         <xsl:if test="lduk_vp_date != preceding-sibling::*[1]/
> lduk_vp_date">
>           <xsl:call-template name="shortdate">
>             <xsl:with-param name="datetime">
>               <xsl:value-of select="lduk_vp_date"/>
>             </xsl:with-param>
>           </xsl:call-template>
>         </xsl:if>
>       </td>
> etc.
>
> This is KIND OF working, but I've realised that my date field
> (lduk_vp_date) contains a time component too, but I'm not interested
> in the time.  My output is therefore looking like this:
> 01/01/2007 00:01:15 Something for january 1st
> 01/01/2007 06:23:05 Something else for january 1st
> 01/01/2007 07:01:11 And another for january 1st
> 05/01/2007 06:45:22 A data row for january 5th
> 05/01/2007 06:45:25 And another for january 5th
>
> Is there anything I can do to avoid this?  I'm restricted to XSL
> Version1 unfortunately.
>
> Thanks
>