Article:		Binary XML, Again
Subject:		Binary Vs GZIP
Date:		2003-08-14 09:15:31
From:		Len Bullard
The VRML community has a long history with GZIP. It even has file name extensions and types to denote that a VRML97 or X3D file is zipped. Because the sizes of the files some years ago were large, zipping became necessary. As bandwidth has improved, it is less necessary but still used. VRML or X3D like most text formats zip well; the bigger problem is as in other formats, images and other non-text media that are used in the multimedia text language. I have quoted some comments from Alan Hudson on why the X3D community considers a binary for X3D to be a must have. See http://www.xml.com/pub/a/2003/08/06/x3d.html Also, along the lines you suggest in your article, the Web3D Consortium has issued an RFP for submissions for binary components and have affirmed their commitment to work with the W3C on this as events warrant. I believe some members of the Web3DC will present on this topic at the upcoming workshop. This question seems to revolve around the utility of a generalized binary for XML. It can be shown that for some applications, a binary is useful not only for performance sake, but for a reason you do not touch on: some customers want opaque content and will not pay for complex content unless they have some reasonable protection against theft by view source. Yes, there are no theft-proof formats from simple binarization, but they still insist on it and contend it is good enough protection.

Previous Message

Next Message

• Binary Vs GZIP
  2003-08-14 23:15:29 Eric Rehm [Reply]
  
  Kendall seems to have completely forgotten about streaming XML applications, e.g., bandwidth sensitive streaming of XML metadata within an MPEG-2 transport stream.

  
  The requirements here have been carefully analyzed by MPEG and resulted in the MPEG-7 TeM (Text Mode0 and BiM (Binary Mode). Both allow a decoder to implement a parser that can skip elements (e.g., for forward compatibility, SAX, etc.) without waiting for receipt of an entire "document" which can be gunziped. Further, MPEG-7 TeM and BiM allows for incremental building and updating of an infoset.

  
  MPEG-7 BiM, of course, allows for compression that meets the broadcaster's requirements to us as little bandwidth as possible for metadata.

  
  Should you scoff at MPEG-7 BiM, note that there is nothing about BiM that is tied to MPEG-7 per se. BiM can be used with any XML Schema.

  
  See www.expway.fr for more info.

  
  /eric rehm
  Singingfish / Thomson
  Seattle, WA
  
  

• Binary Vs GZIP
  2003-08-14 12:29:31 Kendall Clark [Reply]
  
  <u>some customers want opaque content and will not pay for complex content unless they have some reasonable protection against theft by view source.</u>

  
  I didn't touch on this, in truth, because it strikes me as completely lame. I believe you, Len, when say there are people (well, let's be clear, you mean *corporations*) who want such a thing. I for one simply have no interest in that usage of the Web. And, I suggest, the W3C shouldn't get into that business.

  
  As I said, subsets of subsets of various industries are free to define and abide by whatever 'standards' they care to create; what that has to do with the W3C's brief still escapes me.

  
  Thanks for responding, though -- gee, call the W3C a bunch of cultish nutters, as I did last time, and no one says a word. But suggest that the binary-fetishists should steer clear of the W3C (and that it should steer clear of them) and the sharp knives come out! :>
  
  

  • Binary Vs GZIP
    2003-08-16 14:39:23 toto toto [Reply]
    
    i'm a beginner, sot it's may be stupid:

    
    when you want (raster) 2D graphics in a web page, you should use jpeg or gif

    
    so

    
    when you want 3D graphics in a web page, you SHOULD use a binary format (and why not, embeded in textual xml format)