From: Cory H. <ch...@re...> - 2006-10-04 21:30:18
|
Justin Deoliveira wrote: >The patch looks good. A couple of comments: > >1. The result sets created in teh method are never closed. This was like >this before I know, might be a good time to fix it. Not a big issue though. > > done. >2. The pattern matching that is done to find the sequence from the >system tables. Is there any chance it could pick up the wrong sequence >Like for instance if i have two columns in the same table, one named >'foo', and one named 'foobar', that both have a sequence defined on >them, it might be able to pick up the sequence for foobar when searching >for foo. Can we make the comparison a bit tighter? > > Good point... the "last resort" (find the last row) method is a lot safer than possibly getting the wrong sequence. 80% of the time we will get a postgres generated sequence, so let's do a case-insensitive match on "table_column_seq", rather than "%table_column%". Thanks! Cory. |