Thanks for the Friday morning of enlightenment! I agree with Kellyn. I'm a big fan of doing what's right and what makes sense. I don't think storing the entire data dictionary inside the database is the best option. I can see the value for developers that are writing ad-hoc stuff, but those folks (even with comments) should be on the shortest leash and managed correctly in DBRM. (different rant, sorry). Though I may be wrong, I think you should know what you're looking for before you go on a data hunting expedition in a 100TB warehouse. Excellent input from the community, as always. On Fri, Mar 22, 2013 at 10:23 AM, Andy Klock <andy@xxxxxxxxxxxxxxx> wrote: > On Fri, Mar 22, 2013 at 10:14 AM, Mark W. Farnham <mwf@xxxxxxxx> wrote: >> Hmm. Kevlar makes an excellent point. Learning mode on: I think I'm >> revising >> my notion they should flow from the data model to the database. Okay, IF >> you're going to have comments in the database you'll need to make sure it >> is >> a situation that cannot snowball into a performance problem (See Kevlar >> post), AND, they should come from the data model if you're going to have >> them. >> > > > I was hoping that Kellyn was going to mention her case for comments. I've > always been a fan of comments, but was briefly deterred when she described > them being an issue in some cases. Apart from additional space, I thought > they were benign. > > As for a side benefit, I've been known to use table comments (on versions > prior to 11g where dbms_shared_pool.purge doesn't exist) when I have to > invalidate a cursor to force a new hard parse. If you are on a database > with a comment "fast=true", that might very well have been me. > > Andy > > > -- > //www.freelists.org/webpage/oracle-l > > -- //www.freelists.org/webpage/oracle-l