Lists Home |
Date Index |
Been a while since I dropped into an XML-DEV flamewar - XML for data seems to
be doing enough to damn itself these days without needing me pushing it :-)
On Sunday 01 September 2002 2:55 pm, Mike Champion wrote:
> Does anyone really WANT TimBL to insist that every output of the W3C
> match some grand unified vision? ) Sheesh, look at the agony the
> TAG is going through trying to reconcile URIs and RDF-ish/SW-ish
> notions of "identity," REST-ful notions of "resource oriented design",
> and the Web as it really works. There ain't no maps and GPS systems to
> guide people through this wilderness. Posterity will honor those
> who blunder through and survive as "visionaries" but we shouldn't
> kid ourselves that anyone around now knows the True Path and just
> isn't organized enough to explain it to the rest of us.
I see a world of difference betwixt making everything match a grand unified
vision - and internal consistency.
If a group comes up with a universal syntax for representing links, then
goddam every spec should use them unless it's got a good reason. The
advantages of using a consistent syntax to represent links are obvious; for a
start, that would be necessary for the likes of Google to operate in any
hypothetical future where HTML has been phased out by customised XML
vocabularies. Otherwise, Google would need to know all the vocabularies. QED!
But that's just groups building on each other's work, rather than a grand
vision. The grand vision has to be worked out in advance, and changing it is
costly because existing specs built on it might need to be changed. Sod that!
But if groups build upon existing work, then you have a nice incremental
growth that feels around problems as they arise, rather than trying to plan
it all in advance.
Not that there isn't a place for grand universal visions, however. But they
should be guidelines and tools rather than set-in-stone plans. Things like
thinking about interfaces before implementations might be a good grand
universal vision point, since that means that the incremental growth doesn't
get too fragile. For example, since physically retrieving bit strings across
the Internet is handled through URLs these days, it would be possible to
write a protocol that replaces HTTP and the only problem would be getting all
the URL resolver libraries to support it - tiny compared to the problems that
would arise if links were specified like so:
<a server="www.microsoft.com" path="/">Microsoft</a>
You'd need to go round all specs that made references to HTTP servers like
that and alter them to also allow the likes of:
<a phone="+018001231234" path="/">Microsoft</a>
(assuming a wierd "HTTP over raw modem connections" protocol :-)
Getting back on topic, what I'm saying is that the GUV need be little more
than decent engineering practice - the recommendations produced should be
flexible, modular, etc. The rest can be left to evolution.
Oh, pilot of the storm who leaves no trace, Like thoughts inside a dream
Heed the path that led me to that place, Yellow desert screen