[postgis-devel] history management

[George Silva]

Of course I'm interested!

I'll look into it. What version are you using? The one shipped with PostGIS
1.5?

Your idea is really useful and a situation i did not foresee. It may be wise
to have a prefix on every field created.

About the view: unless the user is building a more complicated view, the
table is ready for GIS visualization - and this has a evil side: the user
can edit it. Perhaps it would be wise to publish just a simple view (even
with a underlying table) and updating the geometry_columns for that view
only.

George

On Thu, Feb 25, 2010 at 3:03 PM, strk <strk at keybit.net> wrote:

> On Thu, Feb 25, 2010 at 02:39:22PM -0300, George Silva wrote:
> > Actually it could work in simple tables, but when you are registering
> > history_tables the system updates geometry_columns.
> >
> > I guess that's the only constraint.
>
> And I see it enforces no srid or geometrytype check, which seems useful
> to me I've to say. I'd also drop the messing of geometry_columns as
> the user can always repopulate it later (and hopefuly will be eventually
> turned into a view).
>
> One thing I see is that the _history table is created with *all* fields
> from the original table (using "like <orig>" construct). Wouldn't this
> fail if any field in the original table has the same name of one of
> the "management" fields ? Would it be hard to "escape" the names
> so they have a forced prefix or something to avoid the clash ?
>
> > The proposed version also has functions to turn off "logging", but for
> what
> > i can remember, it doesn't mess with geometry_columns. It's just the
> > registry process that involves a geometry.
>
> I'm not sure I can follow here. I guess I'll make some tests and see
> what happen. If you're interested, a first naive test following
> instructions in the README resulted in this:
>
> NOTICE:  CREATE TABLE / PRIMARY KEY will create implicit index
> "cases_history_pk" for table "cases_history"
> CONTEXT:  SQL statement "CREATE TABLE public.cases_history(history_id
> serial not null,date_added timestamp not null default now(),date_deleted
> timestamp default null,last_operation varchar(30) not null,active_user
> varchar(90) not null default CURRENT_USER,current_version text not null,like
> public.cases,CONSTRAINT cases_history_pk primary key(history_id));"
> PL/pgSQL function "postgis_enable_history" line 66 at EXECUTE statement
> ERROR:  cannot EXECUTE a null querystring
> CONTEXT:  PL/pgSQL function "postgis_enable_history" line 67 at EXECUTE
> statement
>
> --strk;
>
>  ()   Free GIS & Flash consultant/developer
>  /\   http://foo.keybit.net/~strk/services.html<http://foo.keybit.net/%7Estrk/services.html>
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel
>



-- 
George R. C. Silva

Desenvolvimento em GIS
http://blog.geoprocessamento.net
(34) 9664-3717
(34) 8843-3717
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20100225/4459f2e0/attachment.html>