[
Lists Home |
Date Index |
Thread Index
]
Bullard, Claude L (Len) wrote:
>> So all you want is xlink:href as a flag there is a relationship
>> and you will defer any and all other semantic flags to the surrounding
>> markup?
>I could live with a spec that would start with that very minimum and
>later grows to encompass more advanced uses as they appear in the wild,
yes.
That seems to be the same thing Simon is saying: do a bare minimum and
then wait to see what the developers do. I am fond of a strategy that
doesn't standardize too far ahead of fielding; on the other hand, there
was a wealth of experience with hypermedia systems before the web came
along to make everyone start over. We seem to be in a "size of installed
systems vs. proven but not widely used approaches" chinese puzzle.
In other words, XLink works but as Simon says, not enough developers do
that.
Aaggh... who bells the cat?
> Doesn't the http:// tell you there is a relationship?
> What am I missing?
>Doesn't fly too well with relative references does it? ;)
Relative references are harder; good point. You have to sniff with
other rules; so for convenience and reliability, you are back to
href or xlink:href.
Markup is supposed to enable this sharing of constructs. So why
fight it? So far, because no one finds the use case for a common
n-way link to be compelling given that on the web, it is still
just a collection of one way links and there are lots of signs
to denote a collection.
Ok, what does xlink:href buy you that href doesn't?
>And even if
>you get rid of those, you're still really into the
>m/^(http|ftp|file|...):/i heuristics there.
Yes. That is why my first example was generic. You still
have to recognize a URI when you see one. So what you
propose is still the bare minimum needed for a selector.
len
|
