[WF-Protocols] Discussion on Open Model Format

[Chuck Adams]

The spec is at http://www.w3.org/TR/wbxml/ ...  Possibly when originally 
quoted, the phrase "Binary XML" could have been trademarked, since there 
were so many flavors of it, but in general, I think it was a pretty weak 
argument from Veillard.  What wbxml gives you is a reasonably compact data 
stream with a tree structure that doesn't need to be decompressed in order 
to navigate it (unlike a compressed XML stream).  It's not bad for that, 
though I would have much preferred a compression format that understood 
hierarchical data, orthogonal to XML.  But that's a hard problem, and XML 
is in the business of providing crufty and baroque solutions to simple 
problems ;)  WBXML is no exception, and is pocked with all kinds of weird 
cases involving, among other things, codepages (it seems to cater to being 
usable on even 8-bit devices I suppose).

It bears repeating over and over that XML, WBXML, bach, yaml, and many 
other file formats are isomorphic -- they can all express the data model 
equally well.  XML is, overall, for tree-structured data, and quickly gets 
unwieldy for other types of topologies (say, a mesh).  But it's great for 
structuring metadata, and when that's a good fit, then the rest is just a 
matter of which syntactic sugar you prefer.  Focus on what objects you 
want, and what their attributes are, and the format will fall into place.

> Cody Russell wrote:
>
>> On Thu, 2003-07-03 at 01:59, Eason Choo wrote:
>>
>>> This is old, i should've posted my change of vote, i'm all for just 
>>> XML,( and binary xml-if necessary), since xml parsers/writers are very 
>>> easily availbale.
>>>
>>
>> Binary XML is not XML at all.  I talked to Daniel Veillard, author of
>> libxml and former W3C member, and he told me that binary XML is total
>> crap and that you lose all the good things about XML (portability and
>> standardization).  Furthermore, he said "using the term [binary XML]
>> just exposes them to a lot of flack or even trademark infringement".
>>
> Call it what you want but in the end many tools are going to want to use 
> a binary
> file format for compactness and speed of parsing. When I say 'binary XML' 
> I just
> mean a binary representation of the DOM structure instead of the usual 
> ASCII
> representation. He is right that strictly speaking it is not XML. His 
> problems with
> binary are also valid but that doesn't mean that binary formats don't 
> have uses.
>
>>
>>
>> I'd also like to point out that if it is decided that having an
>> extensible format is desired as I was talking about yesterday, this
>> would make this discussed binary format not very useful, whereas a real
>> XML format would work perfectly.
>>
> One can make a binary version of XML just as extensible as the ASCII 
> version.
> There is no real difference other than how the DOM structure is 
> represented
> on disk so you don't loose anything here.
>
> Greetings,
>
>>
>>
>> Cody
>>
>> _______________________________________________
>> Protocols mailing list
>> Protocols at mail.worldforge.org
>> http://mail.worldforge.org/lists/listinfo/protocols
>>
>>
>
>