It is an ongoing debate on whether the fact that an http identifier is
'resolvable' or not is: important; useful; or actually detrimentally
confusing.
I am more or less agnostic. But the general consensus on the web seems
to be moving towards either "important", or "usefully convenient".
But the important thing is, when used as an identifier,
http://id.loc.gov/authorities/sh85090739 is just a string, same as
info:lccn:sh85090739 would be. Doesn't make a difference to a computer
-- except that http is theoretically resolvable, but it works as an
identifier even if it isn't. I don't see any real advantage to using
info:lccn instead -- the extra 15 bytes or whatever in the http one
don't really matter, there's no reason to try and "keep it short".
Some people do see an advantage to using the http one -- I don't
entirely agree with them, but I don't see any _harm_ to it -- except
that it ends up resulting in confusion like this discussion.
But even the http one is just a string identifier with a 'namespace'
prefix, same as the info:lccn one would be, you don't lose anything at
all by using the http one, nor does it give the maintainer any
additional work or obligations.
On 12/6/2010 12:32 PM, Jonathan Leybovich wrote:
On Mon, Dec 6, 2010 at 9:13 AM, Jonathan Rochkind<rochk...@jhu.edu> wrote:
On 12/6/2010 2:27 AM, Bernhard Eversberg wrote:
So, why the trouble to store the entire URI with every record
affected, when the number is all that is actually needed, and
a changed URI most often differs not in the number but in some
other part. For example:
We might have
650 $u http://id.loc.gov/authorities/sh85090739
Only because URI's provide a convenient web standard way to 'namespace'
identifiers. If you only store "sh85090739", what happens if you want to
extend your system to use authorities from more than one source at once?
Maybe you have some LC authorities and some authorities from some other
national library, or, heck, from wikipedia or something. Just "sh85090739"
wouldn't allow that, because the machine couldn't tell what identifier went
with what authority corpus.
I think people are not as far apart on this issue as they think.
While "sh85090739" does not constitute a URI, something like
"lc:sh85090739" would. In fact, the "info:lc" prefix has already been
assigned to the Library of Congress:
http://www.loc.gov/standards/uri/LcInfoURI.html
I would thus support Bernhard's suggestion to use a more compact form.
The URI/URL "http://id.loc.gov/authorities/sh85090739"; (or what if it
were "http://id.loc.gov:9876/authorities/sh85090739"; ? ) has too much
unnecessary system-level information that, while necessary for the
mechanics of retrieval, has nothing to do with creating a unique
identifier. In fact, the Digital Object Identifier (DOI) scheme uses
this more compact form and (in theory at least) lets the user agent
software (web browser, e-book reader, etc.) deal with the mechanics of
network transfer, resource redirects, etc.
