RE: Why is there little usage of XML on the 'visible Web'?

[<juanrgonzaleza@canonicalscience.com>]

Tei said:
>
> I program a quake game engine (telejano.berlios.de). This game engine
> able mods (new games) with this engine. By design decision I recover
> from all errors (as posible), the other options is to crash with any
> error. I, as engine coder, will care about errors, and my users are
> programmers, but my users users are people playing a game. People
> playing a game will get angry if the game crash because a resource file
> is invalid. EVEN if this resource file is XML. So we need
> gratefull degradation. A XML file can define a menu, if the file is
> corrupt, the menu will not show, but the game will not crash. And a
> error mensaje will be at stderr (well...the game console ).
> I guest other engine programmers will feel that this is the right
> decision (Gecko, Microsoft HTML render engine, etc).  But XML, by
> design, has not gratefull degradation. My old movile phone show a
> ERROR 501 (???) if load a invalid xml WML file. So.. its maybe a bad
> idea to send text/xml to the client, because any error => bad error. As
> a engine coder myself I guest HTML fit better than XML for the
> visible web.
>
> I also program remote XUL applications with PHP (here I am the user of a
> engine, Gecko). Because my users will use spanish, and I am a poor
> programer, a PHP code mystake may break my application (XUL will not
> like this solitary &) . So the app will stop with a ugly error.  From a
> intelectual POV, I like XML that way: Your code is invalid, your
> application will break. But not everyone live IT because love IT and
> intelectual questions. So XML will be like ugly for some tipe of
> people ( >99% of the world ), the people that hate ugly errors, and
> compiler warnings, etc.
>
> Example: Warnings are very good for compilers user, but will confuse the
> hell of final users. ( John Sixpack read a alert box with message "The
> buffer has been reread withouth new content, this can mean a
> offset overflow or repeated ticket", his head roll... and die ).
>
>
> Conclusion:
> Why is there little usage of XML on the 'visible Web'?
> Gratefull degradation is good for final users. And is bad, evil and
> greedy for programmers. XML has not gratefull degradation by design. So
> its not good for final users = visible web.
>

The experimental XML website www.canonicalscience.com was being designed
in a full XML way XHTML+MathML+SVG+XSL-FO+XSLT+UnitsML+ThermoML+CML+{a
dozen or more of MLs here} but never completely worked and w3c standards
being really limited (not only Elsevier is using an in-house version of
MathML for instance) whereas scientific MLs are clearly incomplete and not
unified. Then all stuff relaxed to XHTML+MathML+XSLT with SVG ignored and
CSS used for styling.

And next MathML eliminated and after XSLT (due to verbosity inefficiency
in large datuments and incompatibilities even more trivial stylesheets for
publishing are working in Opera but not in Mozilla for instance). And then
Mozilla (Gecko engine) recommends you do not use XHTML if you are not
using MathML, therefore after some months, lot of technical experiences
and checkings, several versions of website, and external advices
(including people from a scientific Search engine will index our site) one
finalizes returning to old but effective HTML + CSS is now being installed
in the server for the definitive version of the website.

And if next HTML5 introduces some sugestions done recently about math then
many of us would find a really good model.

After you continue your research and then understand (as others did) that
something as

<tag1>
<tag2>data</tag2>
</tag1>

is of not great utility off-line (any other format can be so good) and
then you completely abandon the XML hype.

P.S: Your quake-like game looks nice.


Juan R.

Center for CANONICAL |SCIENCE)