[Requests] KML-2.2 suggestion

Sam Halliday sam.halliday at gmail.com
Sat Dec 29 10:04:37 EST 2007

Hi all,

On 29 Dec 2007, at 04:46, Mike Botts wrote:
> While I understand Sam's desires to keep XML "pure" as well as his  
> concerns
> about pushing validation further up the chain (particularly with J2ME
> constraints), I don't agree that we should force coordinate to be  
> wrapped
> with XML element tags. We have done that for large data sets before  
> and have
> suffered the result (very slow parsing, large files, slow transport  
> across
> the web).

I find this to be a very surprising result! Which XML parser were you  
using (I hope it was a "push" parser!)... and are datasets (and code)  
available to show the difference in performance? I'd be very  
interested in reproducing those results. If the difference really is  
as you say it is, I will retract my request. I suppose I could  
simulate the data.... how many data points are we talking about here?

> While parsing a text block of data does require some processing  
> other than
> standard XML software, our experience with developing clients that  
> support
> text blocks has shown that it is neither significantly challenging  
> nor does
> it require much code if the client knows what to expect in the text  
> block.

What have your target platforms been? Have you written for lightweight  
clients such as what is required on J2ME? The range of String  
manipulation libraries in J2ME is particularly restraining... you  
don't even get a 'split' method, which results in bundling an  
implementation. Regex libraries are totally out of the question.



More information about the Requests mailing list