Re: [xml-dev] SGML default attributes.

["Liam R. E. Quin" <liam@w3.org>]

On Sat, 2016-05-07 at 10:29 -0400, Steve Newcomb wrote:
> [...]

> It's interesting that, according to 
> https://en.wikipedia.org/wiki/Formal_Public_Identifier, W3C has never
> registered the owner identifier it uses for itself, "W3C". 

It's easier to check Wikipedia than ISO's registry? Actually I think
the formal registry was hosted by GCA, before they became IdeaAlliance;
you had to pay money and use US mail (not even email). The Wikipedia
entry describes formal public identifiers as a "legacy system",
understandably.

>  If that's 
> true, I guess W3C recognizes only its own notation-universe's
> authority for owner identification, rejecting that of the United
> Nations.


The W3C is made up of volunteers and Member organizations, not only staff. But overall we see decentralization as a goal, and already have a URI/IRI-based name mechanism that works perfectly well as we see it; formal public identifiers are an old syntax that exist in XML primarily for SGML compatibility. It's nothing to do with the United Nations and everything to do with what meets needs. I actually have no idea now but suspect W3C registered itelf.

I actually think it's unfortunate that public identifiers were included in XML; at least one of the people who argued for them so strongly at the time did so (he later told me) because he didn't understand Web architecture at the time and thought of a URI (a URL, at the time) as being like a filename rather than like an identifier with a distributed ownership system.

There were good and well-thought-out aspects to FPIs and a single glaring irreconcilable problem that made them useless - resolving them (in the 1980s and 1990s) involved spending money, sending faxes or US Mail, transcribing data.

But keeping public identifiers turned out to make it harder, in the long run, to bring AFs with us, not technically but culturally, since the concept of architectural forms doesn't depend on a specfic syntax.

The goal of putting SGML on the Web was at odds with defining a clean, decentralized representation for information, because of the need to accommodate legacy architectures. 

Best,

Liam