Steve, * Paragon Corporation (l...@pcorp.us) wrote: > I would almost go with forcing everyone to change their existing > tables to typmod if they want to reap the benefits of PostGIS 2.0, but > most of my clients will not go to 2.0 then.
> What I havn't heard yet is any explanation or description of *why* they wouldn't want to move to 2.0 if it only supported the typmod approach..? See my comment in the trac item - (short answer there are some use cases that just can't be implemented with typmod) http://trac.osgeo.org/postgis/ticket/944#comment:22 > We could try Mark's idea of hacking the catalog tables to > automagically convert table constraints to typmod but that all sounds pretty scary to me. > The upgrade question should not be terribly difficult to solve by just using the current geometry_columns table to help with the generation of the new tables...? New tables -- oh no, you'd have to do an alter (or as Mark wants directly update the catalogs) -- imagine the nightmare if you have foreign key constraints and what not having to create new tables. I have no tables to test with since we don't have typmod, but once Paul gets typmod in there (HINT HINT), I can test with fairly large tables to see how painful doing an alter and dropping constraints would be. > Especially with my special > Inheritcance case -- I can just see that failing miserably and > screwing up my tables. > Can you provide more insight on this..? I'm not familiar with your 'special inheritance case'.. See trac comment. Thanks, Regina http://www.postgis.us _______________________________________________ postgis-users mailing list postgis-users@postgis.refractions.net http://postgis.refractions.net/mailman/listinfo/postgis-users