[postgis-users] Re: if i use postgis, is it true that i'll then have to worry about supporting only 1 database??

Paul Ramsey pramsey at refractions.net
Fri Feb 8 08:43:37 PST 2008

On 8-Feb-08, at 1:39 AM, dnrg wrote:

> ESRI tells me that, at the ArcGIS Desktop release 9.3,
> you'll be able to edit PostGIS data as core
> functionality. No SDE required. This will open doors
> and minds I hope. Paul, any comments on that?

I'll believe it when I see it. Different elements of the ESRI  
marketing apparatus are interpreting the "support PostGIS"  
announcement differently. The most believable story I have heard is  
that ArcServer (nee SDE) will support a PostGIS geometry type, in  
much the same way as it support the Oracle SDO_GEOMETRY type.  So, if  
you think direct editing of SDO_GEOMETRY from ArcGIS (a) works and  
(b) works well then perhaps you have a case to believe that direct  
editing of PostGIS is on the way too.

> Paul, will PostGIS ever have versioning functionality
> for editing workflows similar to ArcSDE? Guess that
> would pervert the data, and then PostGIS would "own"
> the data in a way like ArcSDE does presently. Still,
> many shops find versioning valuable for workflows.

Database lock-in is a fact of life, simple because it is hard to  
migrate databases, no matter how open they are. The best versioning  
solution I have seen is the Oracle implementation, which does  
"workspaces" using the basic MVCC versioning information available  
per-tuple.  I would love to have that, but frankly it requires a lot  
of core PostgreSQL back-end work, and the PgSQL core team doesn't  
have workspaces as a high (or even low) priority item.

If we built a versioning system ala ESRI (side tables and  
references), we could do a somewhat better job, because it would be  
in the database level, not the middle-ware. However, it would have  
the same limitations in terms of requiring client software that knew  
what the heck to do with the stuff.

"Why not."  Explain your use cases that cannot be met any other way.   
There are some, but they are dwarfed by other use cases that have  
higher priority, in my opinion.  I see way more people using PostGIS  
for geoprocessing, for example, so a more robust and faster overlay  
facility seems like an important thing.  Ditto for more core geodetic  



More information about the postgis-users mailing list