[
Lists Home |
Date Index |
Thread Index
]
On 8/19/05, Elliotte Harold <elharo@metalab.unc.edu> wrote:
> The problem may well not be the process and a process based solution may
> not fix it. The simple fact is by its nature, the W3C is spending a lot
> of time in experimental areas. They are blazing new ground, and that's a
> good thing.
It's a good thing if they call it "blazing new ground" (or "leading
the Web to its full potential"), but that's a different thing from
"creating standards." I don't think the W3C has to choose, but they
have to do a better job of labeling. XSD and XQuery and much of the
Semantic Web stuff is blazing new ground. XML itself was carving SGML
best practice into stone. HTML 4 and DOM L1 were basically about
trying to solidify the mud that people were sliding around in. These
offer different values, and calling them all Recommendations obscures
those values.
> Perhaps all we need to change is the attitude and belief that a W3C spec
> cannot be allowed to fail.
I very strongly agree. Some like XML 1.1 (I have finally seen the
light !) were failures from the beginning, others (perhaps DOM is in
this category) have served their purpose and should be allowed to
totter off into the sunset with dignity. Others such as the core XML
specs probably just need some refactoring so that they can be
layered/composed/extended more cleanly. But yes, W3C should learn to
bury its mistakes, retire the expedient hacks, and keep the successes
viable.
> In my experience, if half the community is
> telling you a spec is good, and half is saying it's not, then the half
> deriding the spec is far more likely to be correct and their opinions
> need to be given more weight when deciding whether or not to go to REC.
Hmmm. If half a community is saying a spec is good and the other half
isn't, perhaps they are two communities, one with a use for the spec
and the other without? The web services specs or the semantic web
specs are obvious examples of those that serve the needs of a subset
of the "XML community" but are irrelevant (and somewhat distatesful)
to other subsets. That doesn't mean that they shouldn't advance to
some sort of recommendation status IMHO. At some point you have to
tell the geeks to quit arguing and let the market sort it out.
|
