begin quoting Bob La Quey as of Wed, Feb 20, 2008 at 12:54:44AM -0800:
> On Feb 19, 2008 3:02 PM, Andrew Lentvorski <[EMAIL PROTECTED]> wrote:
[snip]
> > Now, part of the issue is just that the XML specification sucks. The
> > whole distinction between attributes and children was an engineering
> > disaster.
>
> LOL, I made that argument almost a decade ago now. It fell
> stillborne as far to radical a point of view. See
> http://www.xml.com/pub/a/1999/11/sml/
>
I very much like the points made in that article.
I've been tempted, for fun, to write (Yet Another) XML parser that
treated attributes like a convenient shortcut for a leaf node.
That is,
<foo>
<bar>
<baz>
data
</baz>
</bar>
</foo>
could also be written as
<foo>
<bar baz="data"/>
</foo>
...and the resulting parse tree would be the same. Attributes would then
be nothing special. A typographical mechanism of manual compression.
And as I would be implementing this toy parser mostly for fun, I would
have no doctypes or schemas; the promise of XML is that the structure
is consistent enough to allow for parsing; why not take that at face
value and run with it?
Also the requirement that there be just one root node seems...obnoxious.
That makes the process of using XML for streaming data, like log files,
impractical.
--
Yet another nifty project,
But now it's time for bed.
Stewart Stremler
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg
