Re: [PERFORM] Joins on inherited tables

2003-10-03 Thread apb18
OK, so I've had a bit of time to look things over, and appear to be making headway. Here's how things stand right now: I added a function called best_inner_scan used the same way as best_inner_indexscan, but it's a bit more generalized in the sense that it can make append plans comprising of t

Re: [PERFORM] Joins on inherited tables

2003-10-01 Thread Tom Lane
[EMAIL PROTECTED] writes: > So.. does anybody have any advice? Look at set_inherited_rel_pathlist() in allpaths.c --- it forms the best plan for fully scanning the inheritance-tree table. Currently that's the *only* plan considered, and it does not make any use of join clauses. It's possible tha

[PERFORM] Joins on inherited tables

2003-10-01 Thread apb18
Hi, In some situations, it looks like the optimizer does not chose efficient paths for joining inherited tables. For I created a rather trivial formulation to serve as an example. I created the table 'numbers' comprising of the columns id (int) and value (text). I also created the ta