OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] SAX for Binary Encodings (SAD-SAX)

[ Lists Home | Date Index | Thread Index ]

At 3:25 PM -0500 11/8/03, Liam Quin wrote:

>If you choose to pass the string "007" then any interchange
>encoding worth its salt will transmit "007".

Yes, but many people are proposing encodings that aren't worth their salt.

>>  All data in an XML document is text, never anything else.
>All data in a textual XML document is *represented* as text.
>But that text often represents non-textual value.

No, XML all data is text--not represented as text--is text. The text 
may represent something, but it represents different things to 
different people and processes, and what is transmitted is only the 
representation, not the thing itself. When received, the 
representation is decoded into a form that is useful for the 
receiver, which may not be the form the sender intended.

>The value of the textual representation is that it is shared,
>not that it limits all possible data to text.

But all you can share is text, at least in XML. The value of XML is 
that text is much more shareable than binary data.

>Happily, I can refer to dates, to amounts of money, to places,
>to distances, to the colour of my socks, or to the strange
>feeling of dislocation I had when I awoke from a dream last
>night.  None of these things are text, and yet you are reading
>them in this textual email message.

No, I'm reading your representation of them. You have to distinguish 
between the resources and representations. I'm a little surprised to 
hear the W3C XML spec activity lead missing that distinction, but 
perhaps you're not a RESTafarian.

>It's interesting that there was similar opposition to MIME,
>since all email was text, and yet MIME is part of what
>enabled the World Wide Web, and is what lets people send
>each other email attachments such as viruses :-)

MIME, and similar tricks in XML, are a way of encoding byte streams 
into text, but it's still the text that's transmitted.

>The idea that we interchange text is, however, a fiction at
>another level: as far as the computer is concerned, we
>interchange sequences of numbers.  Typically these are
>sequences of octets of binary digits, although there have
>been (and probably still are) systems that use other models.

Yes, but XML is defined at the level of text, not at the level of 
binary data. (See 
http://www.cafeconleche.org/books/effectivexml/chapters/15.html for a 
more complete elaboration of the different layers.) In fact, the 
numbers are just the computer's representation of text. The numbers 
are not fundamental. The text is. I can exchange XML documents by 
printing them out and snail mailing them. I do not have to encode the 
documents as numbers to make them correct or useful.

>When you say that only one view of a layered architecture is
>acceptable, you succumb, I claim, to the hubris of the dogmatic.
>WHen you say that your particular viewpoint is the only one
>acceptable for others, you go beyond the dogmatic to the didactic.

If you want to look at other layers for different local purposes, 
fine. However, I maintain that it's only the syntactic, textual layer 
that's interoperable and interchangeable.

>It might be that interchange of binary information in any
>non-textual representation is not a change we should sanction
>at the W3C (although that will not stop others from doing it),
>but at least let us all make a decision based on careful and
>clear reasoning, and let that reasoning use reproducible
>measurements wherever possible as a basis, not mere dogma.

I'm not sure you're seeing the real disconnect here. It's not merely 
an alternate binary serialization of the infoset that can be encoded 
from text and decoded into text. It's a question of adding new things 
to the binary serialization that do not exist in the infoset, such as 
32-bit ints and IEEE 754 floating point numbers.

When measuring please make sure you measure what counts 
(interoperability and ease of development) and not just what happens 
to be easy to measure (size and speed).  Just because something's 
harder to quantify doesn't mean it's unimportant or less important. I 
claim interoperability is more important than size and speed.


   Elliotte Rusty Harold
   Effective XML (Addison-Wesley, 2003)


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS