Re: How to structure CMS?

[Ean Schuessler <ea...@brainfood.com>]

On Tuesday 22 May 2007 12:16:54 pm Al Byers wrote:
> I think it would be helpful to agree on how content management should be
> used. I am thinking primarily of things like news articles, blogs, forums,
> etc. They all have a lot of similarities and if there were patterns for
> designing them, it would allow for more reuse.
>
> Often with website content there is content and then there is a summary
> piece that may go on a front page and then points to the main content. In
> the past I have used "SUB_CONTENT" as the
> ContentAssoc.contentAssocTypeIdvalue from the publish point to the
> main content. Then I associate the
> summary content to the main content with a
> ContentAssoc.contentAssocTypeIdvalue = "SUMMARY". But I can see that
> there will be situations in which
> things get more complicated, where there will be comments and surveys
> attached to the main content. Also, I recently made the main content hang
> on the summary because at the time it made more sense.
>
> Question #1: Should ContentAssoc.contentAssocTypeId be used to link the
> other types of content to the main content (ie. "COMMENTS", "SURVEYS",
> etc.) or should ContentAssoc.contentAssocPredicateId be used (ex.
> "summarizes", "comments", etc.) and the contentAssocTypeId always be
> SUB_CONTENT?
>
> Question #2: How should multiple page content of the same article be
> structured? All content associated to the publish point as main content
> with incremented ContentAssoc.sequenceNums or should they come off of the
> main content and each page extend from the preceding page?
>
> Question #3: Should we do away with the idea of "main content" and make an
> "article" a complex, virtual content (sort of a placeholder) off of which
> all data content hangs (main, summary, comment, etc.)?

We've spent a little time wrestling with how we would like editing to work. 
I've also dug (a little) into JSR283 and Jackrabbit. One thing I like about 
JSR283 is that it tries to model content as trees of maps with robust 
metadata. We've been doing similar things but on top of CommonsVFS. We bind 
the abstract servlet getResource() functions on top of that CommonsVFS 
backend. Then we can hook in all kinds of different network and local 
filesystems under VFS. Adam and I have definitely talked about binding 
CommonsVFS to the content entities. This would allow servlets (even ones that 
know nothing about OFBiz) to do standard servlet getResource() kinds of 
things and recieve data from Content entities with from/thru date magic and 
all that sort of thing.

-- 
Ean Schuessler, CTO
ean@brainfood.com
214-720-0700 x 315
Brainfood, Inc.
http://www.brainfood.com