[
Lists Home |
Date Index |
Thread Index
]
----- Original Message -----
From: "Al Snell" <alaric@alaric-snell.com>
> On Wed, 19 Dec 2001, Jeff Lowery wrote:
>
> > > The original poster mentioned CSVs which for all intents and
> > > purposes will
> > > always be smaller files than their XML counterparts.
> >
> > Of course, CSV's aren't self-describing. I've spent many hours trying to
> > debug CSV output and the eye-crossover point occurs much sooner with CSV
> > than with XML. Data handles are handy.
>
> CSV can be self describing, if you use the common variant that has column
> headings in the first row :-)
Well ... the 'original poster' was me, so I would insist that
if trying to 'improve' CSV it would result in something
very close to XML.
For example, when column names are 'annotated', but
also must be unique within a schema this is not too much different
from the :
<A property1="some value" property2 = "some value"/>
Comparing this XML with the annotated CSV, XML has
small overhead. However, CSV has no 'arrays' in it , so
after tweaking CSV for 'readability', 'nested arrays' e t.c. -
one may end up with something like :
A property="value" {
B { value }
C { value }
}
e t.c.
Which is almost XML. In fact, just adding </> to XML
would turn it into 'really nice CSV', that would allow
<A>
some text
</>
Of course, then one can easily ruin this CSV with,
namespaces, weird macroprocessor, pseudo-validation
and other useless things, but my point is that XML is
not only the markup language.
I think that if just trying to put CSV on steroids
( readability + arrays ) that could result in something
like XML, so *that* side of XML makes it OK for
serializations.
Sorry if it sounds messy ... XML *is* messy ...
Rgds.Paul.
PS. You know the history behind "qwerty" keyboard?
I mean why the keys are located in that
particular order... Everybody, whom I ask, usually
answers : "by frequency, so we can type faster".
Very ... educational ...
http://www.mwbrooks.com/dvorak/history.html
|
