RE: Server Side Magic.

From: Tomas.Majak@speedytomato.net
Date: Mon, Jun 18 2001

  • Next message: Philipp Hoschka: "[Fwd: [Moderator Action] Server Side Magic.]"

    From: Tomas.Majak@speedytomato.net
    Message-ID: <A8169A4C8016E548B9FA83F181CD1DFD1229DD@stns04.speedytomato.net>
    To: jvierine@mail.niksula.cs.hut.fi, www-mobile@w3.org, www-tv@w3.org
    Date: Mon, 18 Jun 2001 19:06:27 +0200
    Subject: RE: Server Side Magic. 
    
    Hi Juha,
    
    I am not quite sure if I follow you all the way, but it seems to me as if
    you could take another approach;
    
    Firstly you could determine basic information from the client, such as which
    ML it uses. From there moving on to display and memory capabilities. Once
    these are determined you could produce XML according to the memory
    limitations. This XML which now is fitted for the memory capabilities of the
    terminal could then be parsed to the ML that the terminal supports according
    to it's CC/PP. Then continue with tweaking the content of the ML in order to
    make it fit the display, e.g. choose color depth and size of images. 
    
    This could all be done with rules for selecting the content and
    parameterized functions which produce it.
    
    Hope I made my self clear.
    
    Kind regards
    
    Tomas Majak
    
    > -----Original Message-----
    > From: Juha Vierinen [mailto:jvierine@mail.niksula.cs.hut.fi]
    > Sent: den 18 juni 2001 18:21
    > To: www-mobile@w3.org; www-tv@w3.org
    > Subject: Server Side Magic. 
    > 
    > 
    > Hello, 
    > 
    > We are planning to test CC/PP with a web servlet and a browser[1].
    > The CC/PP information can already be sent with the HTTP request in our
    > browser. We also have implemented a servlet stub that reads the CC/PP
    > information and makes it accessible. The problem is with the 
    > server logic.
    > 
    > The server has to do some magic, to come up with a page which fits the
    > client. One way to do this would be to have some kind of data 
    > structure,
    > which contains all the different possibilities to render the page,
    > all of them attached to a CC/PP file containing a profile. 
    > From this it
    > would be possible to some kind of best-fit page to send. 
    > 
    > One simple example would be a site consisting of XML files. 
    > Every XML page
    > has, say three different stylesheets (XSL), which are 
    > designed for three
    > different sized displays. Each stylesheet contains, in CC/PP, 
    >  the size of
    > the page. When a request comes, we calculate, which 
    > stylesheet tranforms
    > the XML file nearest to the clients device. For this 
    > scenario, we would
    > nead every file to contain a reference to a CC/PP file, or 
    > there has to be
    > some data structure describing the site. 
    > 
    > Is there any standardized language for describing a such a 
    > site? There are
    > propably many other ways to do this, which are they. I can 
    > imagine that
    > there has been a lot of talk related to this, but I haven't found
    > anything. Can someone give me a hint on where to look?
    > 
    > 
    > [1] The XSmiles XML Browser http://www.xsmiles.org
    >     + Support for SMIL, SVG, XSL-FO.
    > 
    > 
    > Juha Vierinen - XSmiles.org
    >