[
Lists Home |
Date Index |
Thread Index
]
Costello, Roger L. said:
> Hi Folks,
>
>
>
> In my original post I posed this question:
>
>
>
> Suppose that you are in charge of a Web (you control the funding of all
> the Web sites). Would you issue this mandate to all the Web site
> developers: "All information on the visible Web must be in XML"? If you
> would issue this mandate, why?
>
>
>
> That is, would it be preferable to have a visible Web where web sites
> serve up documents such as this:
>
>
>
> <?xml version="1.0"?>
> <?xml-stylesheet type="text/xsl" href="grocery.xsl"?>
> <grocery-list>
> <fruit>Orange</fruit>
> <meat>Chicken</meat>
> <vegetable>Corn</vegetable>
> </grocery-list>
>
>
>
> Or, would it be preferable to have a visible Web where web sites serve
> up documents such as this:
>
>
>
> <HTML>
> <body>
> <ul>
> <li>Orange</li>
> <li>Chicken</li>
> <li>Corn</li>
> </ul>
> </body>
> </HTML>
>
>
>
> I think that this is a particularly important question, given recent
> discussions.
Yes! because this is highlightinhg the half part of the equation losed
when XML was designed and popularized.
Author --> XML --> User
Most (if not all) emphasis was on XML at the author side whereas the user
abandoned in browsers developers. Does anyone really think that a browser
developer (even one open source) can natively support 200 different XML
specs?
>
>
> Here is my initial stab at the pros and cons.
>
>
>
> An XML/XSLT Based Visible Web
>
>
>
> Advantages
>
>
>
> The burden of converting XML data into HTML is offloaded to the
> clients. Thus, distributed processing is enhanced.
>
>
>
> Customized search engines could be created for specific XML
> vocabularies. Such customized search engines could perform more
> targeted search, e.g., a "grocery-list aware search engine" would
> understand this vocabulary - <grocery-list>, <fruit>, <meat>,
> <vegetable> and would be able to perform targeted searches on grocery
> lists.
This will fail as specific browser fail. Somewhat as i would not wait 300
different XML specific browsers or one with native support for 20
'popular' MLs i would not wait search engines.
>
>
> Disadvantages
>
>
>
> This markup - <grocery-list>, <fruit>, <meat>, <vegetable> has no
> meaning (semantics) to conventional applications (browsers, PDAs,
> cellphones, etc) on the visible Web. Conventional search engines can
> make no sense of the markup. Thus, the ability of conventional search
> engines to index the Web is reduced. The net result is your ability to
> find things is reduced.
>
>
>
> An (X)HTML Based Visible Web
>
>
>
> Advantages
>
>
>
> This markup - <ul>, <li> has definite, universally understood meaning
> (semantics) to a large percentage of applications (browsers, PDAs,
> cellphones, etc) on the visible Web. Search engines can make sense of
> the markup. Thus, the ability of search engines to index the Web is
> enhanced. The net result is your ability to find things is heightened.
>
The problem is that you lose information doing initial XML useless. Then
people would use (X)HTML for the original encoding of the datument.
>
> Disadvantages
>
>
>
> The burden of converting XML data into HTML is placed on the server.
> Thus, distributed processing is reduced.
>
>
>
>
>
> What do you think - should the visible Web become more XML/XSLT based?
>
I think the whole vision is fatally flawed. The web is about comunication.
You need three main 'actors': Author, language, and reader.
Some people wait specific languages, thecnology, browsers and so on. This
is contrary to the web and is really unneded. You can already use OS
tools, specific file formats and specific viewers. You can put a Gaussian
file on the web and information retrieved by user without neeed for a CML
or hypotetical Gaussian-XML reader.
Others wait an all one language. For recent stuff read
[http://www.tbray.org/ongoing/When/200x/2006/01/08/No-New-XML-Languages]
No comments, except the link remember me that number of known XMLs is of
600, not near 200 i asumed previously!
>
> /Roger
>
The solution? I dislike both the xml and the html way. We would find a way
that client can understand any semantics.
For instance why i may choose between <grocery-list> and <ul>. Why one
cannot reuse the <ul> semantics complementing it, instead promoting a new
one. A way is someting as microformats <ul class="grocery">, but there is
lack of support for the semantics on the class attribute, one cannot
combine different classes, data is not unified, etc.
Then the most basic question is not that lightweight technology we can use
for send to client 10^6 lines of code for each non-toy XML-based specific
language. The question is in that generic way i can encode that
information and can be accessed by an unified search engine and can be
understood and rendered by an unified browser.
The advantages of unified framework are inmense. For instance, instead
using SVG for graphics and p-MathML for mathematics one prefer an all SVG
approach. In that case any SVG browser can render math without need for
native supprting MathML.
Juan R.
Center for CANONICAL |SCIENCE)
|
