On Saturday 01 February 2003 05:17 pm, Matt Sergeant wrote:
> On Saturday, Feb 1, 2003, at 19:28 Europe/London, Tod Harter wrote:
> > Well, you CAN simply output RAW XML in a taglibhelper based taglib.
> > Then you
> > can do whatever you want...
> >
> > Personally I do it all the time. The 'automagic' conversion that
> > taglibhelper
> > does from perl data to XML is pretty simple-minded at times. Its nice
> > and 99%
> > of the time its all you need, but it has a few limitations as you
> > found.
> >
> > Of course the danger is that output invalid XML, which will cause your
> > XSP
> > page to crash when it runs, but I honestly haven't found that to be a
> > big
> > problem.
>
> Actually yeah me too. In fact the Wiki uses this technique. At first I
> thought this would be really slow, but libxml is such a fast parser
> that it doesn't seem to make any difference (in fact potentially it
> could be even faster than constructing a DOM piece by piece, because
> everything happens in C space).
>
> Matt.

I had the same concern at first, but I built my entire message base system 
pretty much with raw XML output and, as you suspect, it turns out to be the 
fastest way (though I make no claim that this is true for all possible 
situations, my XML is pretty darn simple stuff). I think your suspicions 
about C space are right on Matt. Perl data structures are VERY handy but they 
are also probably 5 orders of magnitude slower than C code!!! This is 
especially true with large data structures. 

-- 
Tod Harter
Giant Electronic Brain

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to