OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Bad News on IE6 XML Support

* Didier PH Martin wrote:
>Yes true, CC/PP is not implemented yet and therefore the request transaction
>does not include a capabilities record. However, as Cocoon and several other
>server on the market demonstrated, it is possible to recognize the user
>agent by parsing and processing the HTTP header "user-agent". Thus, it is
>possible to recognize the requesting user agent and let the server decide if
>the transformation is to occur on the server side or the client side. So the
>answer is yes. It is possible to extract the information form the HTTP
>header as long as you process it.

This is BAD (i.e. broken as designed) as it was for years.

>Didier said:
>>If yes, the process can occur at the edge or at the client site.
>>Otherwise at the server side. As more client support XSLT transformation,
>>less server load you'll have. Is using a 1 ghz client to perform dummy
>>rendering a best way to use resources?
>Bjorn replied:
>Maybe, maybe not. Why does this matter?
>Didier replies:
>I though you would see why. But again, I just learn for the Nth time how to
>spell the word assume. I'll recap and do some math's.
>a) a transformation takes 1 second to transform a document

If you have a 1GHz machine, maybe. I'm working with a 200 MHz processor,
a transformation certainly takes far more time than 1s. Additionally,
users begin to access the web from a broad range of devices, not only
these kinds of high-end workstations you assume. XSLT requires a lot of
memory and processing power those devices don't offer. And who said, I'd
use 100% CPU time for a laughable job as web browsing?

>b) you get simultaneous 1000 hits on your server. Therefore the processing
>time reserved for transformation is: 1 X 1000 = 1000 seconds.

You want to tell me, transforming the same document 1000 times performs
better than transforming once and delivering 1000 times? Well, if you
don't cache transformation results, you are asking for trouble, but
that's *your* fault, since *your* design is broken.

>c) if the transformation occurs on the server side, this same server just
>got its workload increased especially if the processed XML document is
>dynamically produced by a server side script.

I'm sorry but delivering the same XSLT transformation sheet to 1000
clients takes far more resources than one simple transformation.
Dynamically generated documents don't count; those scripts should have
produced XHTML instead of something else.
Björn Höhrmann { mailto:bjoern@hoehrmann.de } http://www.bjoernsworld.de
am Badedeich 7 } Telefon: +49(0)4667/981028 { http://bjoern.hoehrmann.de
25899 Dagebüll { PGP Pub. KeyID: 0xA4357E78 } http://www.learn.to/quote/