Hi,
> From: Phil Barker [mailto:[log in to unmask]]
>
> Paschoud,J wrote:
>
> > 1) Somebody's nth rule about data normalisation (etc, etc)
> deprecates
> > attempts to embed extra information (such as a date - relative or
> > absolute) within data that purports to be something else (such as a
> > version identifier). If a human or machine recipient of 2 or more
> > files/objects wants to determine which is newer, shouldn't
> s/he/it be
> > comparing some separate datestamp attributes; not version
> identifiers?
> Well, newer does not always mean most up to date, for example:
> Mozilla 1.0.2 - Released January 07, 2003
> Mozilla 1.1 - Released August 26, 2002
> Mozilla 1.2 - Released November 26, 2002
>
> > If so, hash-derived v-ids are fine, because it's only their
> sameness or
> > difference that matters, not any knid of ordering of them.
>
> But then you might as well just have identifiers for the HTML
> files and not
> bother about lifecycle.version.
Unless I'm missing something here, with hashes as identifiers, your
reference to a resource breaks when you change the resource file (and it's
hash).
I'm thinking that a resource may evolve over time, but it's main
use/relevance will stay the same. So surely we do need both - the identifier
to say "this is a particular resource", and the lifecycle.version to say
"this is version 1" (or whatever value), otherwise how will we be able to
update.
> > HTML viewable/usable? Does each viewing of the HTML
> constitute a new
> > 'version' (I hope not!).
Yikes! I'll definitely not be going there.
> > I think this produces extra reasons to support
> > Paul's suggestion, that only the original XML content should be
> > versioned, not the result of any combination of processes
> to render it,
> > because not all of those processes may be under the control
> of whoever
> > cares about the versions.
> >
>
> Yes, I think I agree with this. I think this relates to a
> question Andy
> Powell has asked on a couple of occasions: what is it that
> identifiers in a
> LOM record identify, i.e. what is a LOM record about? He
> phrases this in
> terms of the entities given in the Functional Requirements for
> Bibliographic Records
> (http://www.ifla.org/VII/s13/frbr/frbr1.htm#3.2).
> I think XML and HTML are two different manifestations of the
> same work, in
> two different formats, but not necessarily two different versions. You
> could encode the relation in the metadata using isFormatOf
> and hasFormat
> relationships; if the XSLT does more than re-format, then in
> the LOM you
> have isBasedOn/isBasisFor reltationships--for that to work
> the LOM records
> would have to be about the manifestation, which makes sense when you
> consider the technical category as well.
I need to find more time to get my head around this stuff :-/
Cheers,
Dave.
|