Lists Home |
Date Index |
I don't know. We did some work with it on
IDE/AS and it wasn't that horrific but one
did have to control the renderer because
essentially, passing in a locator and having
a common API aren't that different. It's still
just a message with an implied semantic.
The HyTime guys expended a lot of paper pulp on
it, and the Hyminder engines took the idea seriously.
The idea that locators came down to handful of basic
types seemed worthy.
If all one is passing is the locator type plus
the values defined for it, why is that flawed
(other than the a priori knowledge required)?
I am familiar with the fragility issues. Even
the Hytime guys said that robustness would vary
for all the usual reasons.
From: Joe English [mailto:email@example.com]
Bullard, Claude L (Len) wrote:
> I think this is at least the fourth time
> over the last 12 years this inescapable
> issue of having common link and locator
> types that can be interpreted correctly
> by different format handlers has come up
> in my field of observation.
Has anyone ever gotten it to work?
Just curious. I've always believed that this whole
idea was fundamentally flawed (and that HTTP's
content negotiation scheme is basically broken too.)