Forum OpenACS Development: Response to New implementation of tree_sortkey

Collapse
Posted by Don Baccus on
Oh, I've checked the thread and David's "create or replace" hack written in PL/Tcl.  Cute!  I may look into this later.

Frankly I'm not too worried about the speed of finding parents at this point.  For the most part it's done on limited sets of objects, i.e. site nodes, bookmarks, and the like.  The tree_sortkey index can be used to diminish the rows scanned somewhat for larger structures and these limited sets usually aren't going to be large anyway.

So I'm probably willing at this point to put aside a faster solution until we're ready to require PG 7.2 or at least until our second release.  Note it shouldn't require any change in the datamodel's representation of sortkeys so we can make such changes transparent to OpenACS 4 sites in existence when we make this further improvement.