I think "not necessarily" is the operative word. I also like the term
convoluted to describe it. But I will try to describe it better from the
top. The overarching structure is a TC which will have projects and
documents. For instance
TC 08 has
Doc 201
Doc 202
Doc 203
Doc 204
Proj1 = combined revision of Doc 201 and 202 (i.e., project is associated
with two documents)
Sample activities:
Act1 = first draft for Proj1 (activity associated with project)
Act2 = inquiry about creating project for revision of Doc 203 (activity
associated with document)
On top of this, take into account an activity like
Act3 = meeting for Doc 210 (does not fall into a TC)
Now imagine a scenario where Act4 = is a first draft on a new Proj2 which is
a revision of Doc 203 (activity associated with a project).
If I wanted to do a search for all activities for Doc 203 I have instances
where it was not associated with a project and then later had a project.
I have thought about creating a fake project Proj00 which would house all
documents not associated with a TC (would have to make up fake TC) or a
project. But, would I then be able to extract activities associated with Doc
203 if it once belonged to Proj00(along with many others) and now is in
Proj2? The problem is that, for instance, any activites associated with Proj1
(combined revision of Doc 201 and 202) will able to BOTH documents. But
activities associated with Proj00 will not apply to all documents within it.
Does this make more sense? It is a spiderweb. I have inherited this mess
without much forethought to feasibility of design. So, I am doing my best!
Thanks again!
Michelle
> If one takes everything that you said and implied literally you end up with
> a spiderweb of relationships, with most of them being "not necessarily"
[quoted text clipped - 54 lines]
> >
> > Thanks!
Fred - 29 Apr 2008 22:26 GMT
Hello Michelle,
On my first pass at rigorously absorbing what yopu wrote my brain started to
explode so I had to shut it off. I'll try to try again, but in the
meantime, a few thoughts.
Even a spiderweb can be created / documented in Access relationships.
Write a query that links whatever tables when needed. And even the most
complex scenerios (like "sometimes many-to-many") can be accomodated with
itermediate linking tables. I'm just not sure how informative / useful such
a spiderweb will be. If you have no rules (= all exceptions, fluid
definitions and infinite linking possibilities) then you don't have data
structure that serves you.
Not sure to what extent those terms are part of the real world ( = you
probably shouldn't / can't change them) vs. inventions for the DB
implementation (= you can change / redefine / reorganize them.)
My gut feel is that when you have a multi-tier structure that the core
structure should follow one attribute/nature of entity, and then other
types of tabled-items should be just "hung on" to the main structure. I'm
guessing that your core structure should be big and small elements of
objectives for work or work (projects, tasks, activities etc. ). This would
be a pyramid of 2 or 3 tables where the records are strictly this. And that
the other entitities (TC's & pieces of paper) should be merely attached to
this core structure. You'd probably to get rid of a current double meaning
of "document" (something to accomplish[e.g. completion of a piece of paper],
and a piece of paper) or split it into those two.
Just a few thoughts.