RE: The SAX2 standard ? (was SAX2 bugs -- please file!)

[Nicolas LEHUEN <nicolas.lehuen@ubicco.com>]

I reckon that I may have been a little bit too enthusiastic about the
"standard" word. Moreover, I mixed together two different problems :
standardization of XML and standardization of XML processing APIs.

"Standards", or "Specs", or "Recommandations", or "Requests For Comments",
you name it, are a way to answer differents concerns.

My concern, as an XML user, is to have the guarantee that the XML content I
produce or get can be handled by my tools and applications. This is why I'm
worried about the recent Blueberry debate, that made me ask myself "how
decides, anyway ?".

My concern, as a XML developer, is to be able to easily integrate together
tools and libraries and weave my application code in between. This requires
sound APIs recognized by the industry. If such an API doesn't exist, then
I'll have to write adaptation code - annoying, but no big deal. However,
with a very simple event-based API, you have a great integration power, so
it would be very sad not to have a sound and recognized API for this. This
is why I wanted to know who was making decisions on the future of SAX.

My concern, as a tool or library developer, is to give my customers tools or
libraries they can use to write their application. I have to make sure that
my products integrate with as most other tools as possible. I can't invent a
new XML processing API every morning, because my customers won't appreciate
wasting time and money to write adaptation code. Here, XML manipulation APIs
are vital. And I'm not alone in this case. So today, it seems that one of
these vital API is written and maintained by benevolent people, and I'm
worried about how the industry, standardization organisms and so on interact
with these people.

It's quite fascinating to see all those XML technologies emerge from a so
wide variety of people, organisms, companies, etc. The Internet was also
created from a common initiative. But AFAIK, RFC management has been
centralized early on, and it seems to me that some parts of the current XML
specification efforts are much more complicated than most RFCs. How are we
going to make sure that all these specs and technologies keep coherent with
each other ? How are we going to prevent XML from exploding into multiple
sub-specs, thus failing to reach the possibilities of XML as an unifying
data exchange system ? I just hope someone knows how.

Regards,
Nicolas

>-----Message d'origine-----
>De : Bullard, Claude L (Len) [mailto:clbullar@ingr.com]
>Envoyé : mardi 31 juillet 2001 21:43
>À : Nicolas LEHUEN
>Cc : 'xml-dev@lists.xml.org'
>Objet : RE: The SAX2 standard ? (was SAX2 bugs -- please file!)
>
>
>From: Nicolas LEHUEN [mailto:nicolas.lehuen@ubicco.com]
>
>Let's do this in order with some points up front.  Standards are 
>legal documents.  Specifications are design documents.  Despite 
>what you read, think or want to believe, the W3C is not a standards 
>producing organization.  There are good reasons for that and the 
>more people push the W3C to be that for whatever reason, the more
>they weaken its ability to incubate technology.  Caveat vendor.
>
>>1)  We really, really need a standard for this part. SAX is 
>the de-facto
>one for the moment.
>
>SAX is a specification.  What would change if it were a standard?  What
>would 
>be improved by any standards process?
>
>>2) Yet, DOM is an "official" spec supported by the W3C,  
>whereas SAX is
>not,
>>and gets to live on SourceForge. What will be the 
>standardizing power of
>the
>>SAX community ? 
>
>None.  Governments ratify standards.
>
>>I mean, what guarantees that a vendor won't change the API
>>to suit its own need, thus breaking all possibilities of tool 
>chaining ?
>
>Nothing guarantees that with a specification or a standard.  
>Conformance 
>testing proves the case, but there is no enforcement other 
>than the right 
>to attach certain signs to the documents claiming such.  Nothing.
>
>>If SAX is not a standard, what are we left with ? 
>
>Precisely what you have today or implement tomorrow.
>
>>How can we have high-level standards such as XML
>>Schema, SOAP, etc. without having a standard for the very 
>core building
>>block that SAX is ?
>
>By implementing or purchasing code that passes conformance tests.
>
>>3) SourceForge is neutral, OK, but is it the purpose of the people at
>>SourceForge to host standards ? Can they accept the responsibility of
>>hosting such projects that may be central to the industry in 
>the years to
>>come ?
>
>No.  They shouldn't.  This does point out the question one asks of the 
>hosting authority:  show me the process that supports the lifecycle.
>
>>4) Note the "official" in question 2. I'm very far from being 
>an expert in
>>standardization organisms, so please pardon me if I make mistakes, but
>>shouldn't XML be an ISO / IEEE / other standard ? 
>
>Just as ISO has a version of HTML, ISO could have a version of XML but 
>since it already has SGML, and XML is a conforming subset, not much 
>would change.  Independent consultants such as Tim Bray have vocally 
>opposed XML standardization and it is not likely this issue will be 
>brought up again for awhile.  Meanwhile, XML remains a product of 
>the W3C and a conformant subset of ISO 8879 SGML.  This relationship 
>is sufficient to ensure stability of the product.
>
>>What are the historical reasons that gave the W3C control 
>over the XML 
>>standardization ? 
>
>A self-selected group of SGML experts chose to create a 
>specification for 
>the W3C product and solicited and gained sufficient support from the 
>SGML community, ISO and the W3C to create the XML specification.  
>
>>What is the position of the
>>"traditional" standardization organisms regarding W3C ? Does 
>the industry
>>consider the W3C as being pertinent for these issues ?
>
>There are liaison relationships between standards 
>organizations and the 
>W3C.  Industry for the most part considers these pertinent and 
>adequate. 
>Some government standards bodies create standards by 
>reference; that is, 
>they create a standards document that references the W3C 
>specifications 
>are being authoritative for the technical information requirement to 
>meet the standard's requirements.  The government body retains the 
>authoritative position for the standard itself.
>
>>5) I'm don't know much about SGML. What is the 
>standardization level of
>this
>>technology ? Is it ISO, IEEE or something else ?
>
>ISO 8879.
>
>>6) learning that SAX is "only" a de facto standard hosted by 
>benevolent
>people is quite 
>>frightening.
>
>Perhaps.  But so far so good. 
>
>>7)  What worries me here is the fact that the W3C is quite 
>silent about
>SAX. 
>>What does this mean ?  
>
>It is tacit acknowledgement that the W3C need play no active role in 
>the specification of SAX.  Since there is considerable overlap among 
>W3C, OASIS, ISO and independent members, there is considerable 
>expertise 
>among all of these groups on SAX.
>
>>What kind of standardization dynamic does it expose ?
>
>It exposes the fact that SAX is not a standard.  It exposes the lack 
>of an immediate and pressing need to change that.
>
>Len Bullard
>Intergraph Public Safety
>clbullar@ingr.com
>http://www.mp3.com/LenBullard
>
>Ekam sat.h, Vipraah bahudhaa vadanti.
>Daamyata. Datta. Dayadhvam.h
>