[
Lists Home |
Date Index |
Thread Index
]
Looks like we really are going to do a navel introspection
thread. Ok, it's late summer here in the northern
hemisphere and you guys down south can use these
threads for kindling....
From: Didier PH Martin [mailto:martind@netfolder.com]
>I was referring to elements used as links. I want to be sure to
>understand you well. Are you saying that links are not necessarily
>associated to operations (1) or that in no case it should be associated
>to an operation.
One. It isn't "necessarily" associated to an operation, although
I am one who sees a link and thinks function call.
>Perhaps a link is at its more abstract level simply indicates a
>relationship and when mapped to an element,
It can, but a link like that is a named set of properties.
> this latter add some
>implicit or explicit operation responding to the question an interpreter
>may ask (2): What do I do with this link.
If you want to link from HTML or any other source into a target system,
how does the target know what to do with the values you pass regards
locating the anchor? It depends on the native structures of the
target, yes? In other words, CGM has to know what it means by
"circle" and what to do if that is requested. It is no good
to simply say, that is a representation or a circle, because the
processor has to actually locate something in the bit bucket. Hytime
proposed a set of locator types based on the kinds of structures
one could anticipate. In a sense, that violates encapsulation because
the locator type has to be configured in advance and that makes it
difficult to say, "this is an abstract resource" and let the server
sort it out from the headers and the registered types. Still, it
illustrates some of the problems with a common addressing system
based on absractions of type (and why the new Office XML system
from MS will be really *entertaining* in the first release).
> So to speak, a kind of href
>with, as value, a locator or a name that could be resolved into a
>location in addition to other attributes (including the element's name)
>that help the interpreter to know what to do with that link.
In the dim past while I was off doing other things, the idea that a
link should if operational, name the notation processor of its
target got lost. Mime types were supposed to do all of that.
(1) I agree on the abstract level since I may indicated just a
relationship to be interpreted with one or more semantics dependent on
the interpreter.
(2) You know interpreters are like 4 years old kids, they ask a lot of
question and do not clean their mess :-) :-)
Sure but a messy house is a sign that one has kids. A smelly house
is a sign one has babies. A clean spotless odor free house is
a house on the market. That's life. Read the signs.
len
|
