Subject: Re: [xsl] rephrased: passing parameters to generic templates From: "andrew welch" <andrew.j.welch@xxxxxxxxx> Date: Wed, 12 Apr 2006 17:32:15 +0100 |
On 4/12/06, Wendell Piez <wapiez@xxxxxxxxxxxxxxxx> wrote: > At 07:59 AM 4/12/2006, Andrew wrote: > >If I were using 2.0 I would use a tunnelled parameter that was created > >at the <para> matching template and then read in the <a> matching > >template, but the OP didnt day whether they could use 2.0 or not.... > > What would the advantage of this be over the old-fashioned way? > > Would we save a few microseconds? > > Isn't the direct approach easier to maintain? I always try and avoid ancestor and descendant - whenever I use them my conscience tells me I'm being lazy, and makes me use a different approach... In this example there are many <a>'s to one <docBase>, so traversing ancestor each time to retrieve the same value seems like a waste - if there are 3 <a>'s, that means 3 trips up the same ancestor axis to get the same value (I guess it probably gets optimized, but why rely on that?) much better to store the value in a tunnelled parameter.
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
RE: [xsl] rephrased: passing parame, Michael Kay | Thread | RE: [xsl] rephrased: passing parame, Michael Kay |
RE: [xsl] rephrased: passing parame, Florent Georges | Date | RE: [xsl] xslt 1.0 namespace on out, Michael Kay |
Month |