Hi,

I just wanted to make a few minor points on this
thread.  I personally like XML, but given how oversold
most technical approaches are, it is important to look
at any drawbacks and characterize them (so I am glad
Rich brought up the issue).

For transfering data between systems, XML does provide
a uniform way to do it.  Comma separated file type
approaches also will give a certain degree of
uniformity, but less than XML because of validation
through dtds/xml schema.  The drawback of XML would
probably be the file size.  For simple columns of
numbers you could have overheads associated with tags
that increase file size by a factor of 2-5.  This
would also slow the parsing of the file by roughly the
same factor.  For small files that may not be a
problem.  But if the file size gets large, it could
be.  Also if the file size gets large you will have to
use SAX (sequential access) instead of DOM (random
access, file is read into memory).  SAX will be a
little harder to implement than DOM for the same
system.  Comma separated files would not have this
kind of problem, since each line is read and entered
into the system receiving the data.  Ultimately if you
are willing to pay the overhead you can use XML.

It might be nice if XML had a structure to reduce the
number of tags it requires.  This could solve the file
size expansion that arises from the tag names. Let's
say you had a system that does records the temperature
as a function of time.  So you have times and
temperatures in the comma separated file.  The XML
file would look like:

<tempdatapoint>
<time>
12
</time>
<temperature>
45
</temperature>
</tempdatapoint>
... even more <tempdatapoint> elements ...


A single file would have many <tempdatapoint>
elements.  You might want to use attributes for this
data, but you would still have a bunch of extra tags. 
If there was a structure in XML that allowed this
information to be more compact it could solve the
problem.  Something like:

<flatelementset rowname="tempdatapoint" delimeter=","
linedelimeter="\n">
<element number="1" name="time">
<element number="2" name="temperature">
10,45
11,46
12,52
...
</flatelementset>

The XML parsers like SAX and DOM would then have to be
smart enough to expand the second compact notation
into the first notation when it is read by a program. 
If anyone know whether this exists, please respond. 
BTW, I mean in general not as a proprietary extension.
 But it would be nice to know even if it was just a
proprietary extension.  Much more thought would have
to be given to this kind of an extension.


At the same time, XML does have some advantages over
comma separated file type approaches.  Let's say that
the schema of the data you are recieving does not
match well to the system you are entering it into. 
Maybe the comma separated file data is not normalized
where the database is.  If you are using XML, it is
nice to use XSLT to transform the data into a schema
that is more similar to the database before it is
entered.  I think this would be much easier than
writing a parser to do the same thing.  Also sometimes
you may be recieving more data than you really need to
enter in your database.  XSLT could also be used to
filter the data before it is read by your parsing
program.  

XML might also be better if you had a richer data
structure (more like trees than rows in a database). 
I know that trees can be emulated using rows, but it
might be a little harder.  I took the case I did
simply because it represented the strongest objection
to using XML approaches.

For web applications and for any other applications,
there are always many ways to achieve the same effect.
 And it is can be very application dependent which
technology you want to use.  I'll just say that I like
having the option of using SOAP/cocoon approaches
available.  

Bill Gooding




--- "Richard O. Hammer" <[EMAIL PROTECTED]>
wrote:
> Dave Reid wrote:
> > The major point is ease of interchange between
> systems. I'm working on a megaproject with 25
> interfaces all to different types of systems.
> Without XML as the consistent underlying standard
> for all transaction interchange our work would have
> been much harder, and more costly.
> 
> Dave,
> 
> I wonder if the advantage you mention here has to do
> with the fact 
> that XML is text.  Could you interchange data
> between 25 interfaces if 
> the specification required ASCII or ISO-8859-1, with
> the same ease 
> that you achieve with XML?  If so, then I still do
> not see the 
> advantage of XML.
> 
> Rich
> 
> 
> _______________________________________________
> Juglist mailing list
> [EMAIL PROTECTED]
>
http://trijug.org/mailman/listinfo/juglist_trijug.org


__________________________________________________
Do you Yahoo!?
Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop!
http://platinum.yahoo.com

_______________________________________________
Juglist mailing list
[EMAIL PROTECTED]
http://trijug.org/mailman/listinfo/juglist_trijug.org

Reply via email to