What's wrong with the current ConceptualModel?
Any model which does the following appears broken:-
-
Forces people to use anything more than a GUID for entries.
-
Constrains entries to modelling current weblogging trends without being 'blog agnostic.
-
Confuses documents/articles/entries with the publication events associated with the entries.
We have a real chance here to make something that is supremely simple, totally trend-free, and unquestionably usable. It should be about simplicity, ground up design, and precise semantics. If it's not, then it'll just continue to be replaced, ever more confusingly, for users.
For example, an entry/article/document can be published many times in many locations, can be updated, and can go through numerous other processes. If entries became a much more general thing, independent of when they were published, that would be a helpful level of generalization.
Therefore, we propose here...
The Revised Conceptual Model
The key distinguishing features of this model are:-
-
Entries are defined very broadly, so that they may be articles, documents, images, stylesheets, templates, FOAF data, etc. (@@ Perhaps rename them documents).
-
The only required property of an entry is GUID, so that people can build their entries with only the metadata that they require.
-
Events, entries, and fora are pairwise disjoint (fully distinct from one another). Events are properties of fora.
An entry in this model is defined as follows:-
-
Only one madatory property: GUID.
-
A core set of properties: URI*, author*, title*, creationDate?, description*, content?. title may be repeated due to internationalization.
-
Extensible to include any other proprties that people see fit, e.g. modificationDate, version, derivedFrom, etc.
That's it; the idea is very simple. The rationale is more complex. Starting from the ground upwards means that things can be very simple at the base (the smallest entry possible is just <entry guid="someguid"/>), gradually getting much more granular.
Entry/document extensibility: Any other properties must specify what their maximum occurance is: for example, entries can only have one creation date, but multiple authors. This project should define a core set of properties for use on entries. Note that authors of new properties should be very careful to describe the exact semantics, and take into account internationalization, etc. For example, in the core set of properties, title should have an unbounded amount of maximum occurances since one can represent titles in different languages.
@@ Issue: is a URI a permalink for the entry, or for the entry's content?
An event in this model is defined as follows:-
-
Must be associated with exactly one forum, and one entry.
-
Must/should include a publication date.
Again: events are associated with fora, not entries.
Benefits of the New Model
This opens entry up to being able to describe stylesheets, templates, images, articles hosted by other people without fear of copyright infringement (sorta like Xanadu's transclusion), and so forth. It's flexible. It's easy to understand. It's simple.
See also the use case on ConceptualModelRevisited.
What could it look like?
@@ example serialization!
FAQ
@@ got questions/comments? bung them in here, or ConceptualModelRevisited if they're more general.