[GRASS5] Re: Vector format proposal

[Eric G. Miller]

Well, here's my two cents on data models.

I fully agree with David regarding graphic vs. topological
representations of "entities".  I've seen plenty of shapefiles with
screwed up geometry to know this is a real problem.  But, I'll grant the
topological "coverage" concept is a bit more complex to manage vs. the
graphic representation.

In regards to attribute tables: I would think a one to one mapping from
each map entity to an attribute table via a unique integer key is the
best solution.  So, for each AREA, LINE, SITE, LABEL, NODE, etc there is
a table.  It might be, we want to maintain some separation between
internal numberings and what the user uses for a key.  Otherwise we
can't change a numbering after an edit/rebuild cycle because the linkage
with the attributes will be broken.  A simple two integer table is all
that is needed, the user is only presented with setting the "user_id"
whereas the other is maintained by the system.  Example:

AREA_ID | USER_ID
      1 |       1
      2 |       1
      3 |       2
      4 |       1
 ...

So, when v.digit asks for a "category number" it sets the "USER_ID" with
whatever the user wants, but the "AREA_ID" is incremented as the next
largest value (for that table).  A rebuild might renumber the "AREA_ID"
to make them sequential, but the "USER_ID" is never touched.  Thus,
linkage to external tables always works.  I think I'm getting redundant
here ;)

-- 
Eric G. Miller <egm2 at jps.net>

---------------------------------------- 
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo at geog.uni-hannover.de with
subject 'unsubscribe grass5'