In article <Pi********************************@kosh.hut.fi> ,
Mikko Nummelin <mn******@EI.cc.ROSKAA.hut.fi> wrote:
[About the convention of using http: urls for name spaces]
% Sorry for my stupidity. IMO, using variable or file names of that type is
% extremely misleading and comparable to a malpractice of naming the further
% Linux kernel gzip packets in a way like
%
http://www.microsoft.com/2.6.1
%
% /dev/null/2.4.27heydude
%
% , which are, well, valid UNIX file names, but ...
The problem is that you need to have a globally unique identifier. What
would you like to use, that you can guarantee nobody else will use, or
at least you can claim they're not being reasonable if they do use it?
People started using http: protocol URIs because the network address
is guaranteed to be unique and people can reasonably expect to
control the URIs associated with addresses that they `own'.
Someone could have defined a protocol for name spaces, like
ns://www.w3.org/1999/XSL/Transform
and saved everyone a lot of trouble, because the convention of putting
documentation at the web address used for name-spaces only arose _after_
lots and lots of people just assumed they'd find documentation there,
and raised a fuss.
Anyway, I think it's more confusing than misleading (certainly not
`extremely misleading'. I have no idea what your analogy is supposed
to mean (it's also confusing), but it seems likely that it's not
particularly apt. Nothing has been mislabled, there's just a URI
being used in a context which doesn't suggest there are any resources
assoicated with it, and which has no resources associated with it.
--
Patrick TJ McPhee
East York Canada
pt**@interlog.com