Matthew A. Schneider wrote: (01)
> So a link to a particular version will remain intact? I agree that
> versions should get new number. (02)
Yes, a link to a version of the paragraph would be direct.
It does raise the issue of how to tell when one paragraph is a new
version of another when using just transclusion fragments. E.g. if I
want to indicate to a transcluder that newer version is available than
the one they are looking at, just at the paragraph level.
That might require strategies to supply the metadata, like maintaining
the references between paragraph changes on the server side. (03)
I guess the view I'm espousing is that the links (locators) for a unit
of stuff should exist along the time/version dimension, not that the
time/version dimension exist over a single link.
I.e.
Thing -- Link 0
Date -- Date 0
Thing + Date 1 -- Link 1
Thing + Date 2 -- Link 2 (04)
...rather than,
Thing -- Link 0?date=1
Thing -- Link 0?date=2 (05)
In the former that time/version dimension is a property of system
connections. In the second, it relies on traditional date syntax and
user semantics and there's no composition to suggest that a Thing is
part of the deal before the link is resolved. (06)
This approach is just a working hypothesis for me at the moment. I'm not
sure it is possible to expose it fully in URIs, and it is up for debate
and refutation. (07)
Comments welcomed, as always. (08)
--
Peter (09)
--
This message is archived at: (010)
http://collab.blueoxen.net/forums/cgi-bin/mesg.cgi?a=tools-yak&i=4258FFFF.8090307@concept67.net (011)
|