Return-Path: pg_adm@postgres.berkeley.edu Received: by postgres.Berkeley.EDU (5.61/1.29) id AA00209; Wed, 24 Jun 92 03:37:35 -0700 Message-Id: <9206241037.AA00209@postgres.Berkeley.EDU> From: "Schenkelaars.V.F" Subject: R-tree problem (the continuing story) To: postgres@postgres.berkeley.edu Sender: pg_adm@postgres.berkeley.edu To: mao@postgres.berkeley.edu (Mike Olson) Date: Wed, 24 Jun 92 12:41:25 MET DST Cc: postgres@postgres.berkeley.edu (Postgres mailing list) X-Mailer: ELM [version 2.3 PL11] Dear Mike, > what happens is that your intersection routine returns NULL if two boxes > don't intersect. the quadratic-cost split algorithm from guttman's paper, > as i implemented it for postgres, computes the sizes of the union and > intersection of the n-dimensional boxes in order to choose an optimal > node split. Size3Box() got passed the null pointer from the intersect > routine. > > i'll fix the split code to check for this condition before calling > the intersect procedure. for the time being, the change described above > will solve your problem. i successfully ran your test program on my > system this afternoon, with these modifications. > > incidentally, i'm pleased to see you folks are extending the access method > classes like this. your code was clean, and you uncovered a genuine bug. > let me know if you have further problems -- response times should get > shorter, now that summer is here. > mike olson > project sequoia 2000 > uc berkeley > mao@cs.berkeley.edu This is very nice, but there is still a problem left. I've already asked this before but I didn't get any answers. When I define a index first and than add all the tuples, everything is going fine. But when I first add all the tuples and create an index afterwards, the size of the postgres backend grows extremely and finally terminates with a core dump. What is going wrong. About adding an new access class, how do I start with doing so? I have to implement an R-tree based index structure so I "only" have to change your R-tree code somewhat. A brute solution would be changing the R-tree code and recompile Postgres. Howerver I am wondering if there is not an elegant solution for it. Which files are dealing with the R-tree? Regards ____________________ \ / . _ _ _ _ | \/ ( ( \ (_ (-' ( \ | -------------------------------------------------------------------------------- Vincent F. Schenkelaars | Company: FEL-TNO | Private: Email : V.F.Schenkelaars@tnofel.fel.tno.nl | address: Oude Waalsdorperweg 63 | Groenhovenstraat 12 Postbus 96864 | 2509 JG Den Haag | 2311 BT Leiden The Netherlands | The Netherlands Phone : Holland 070-3264221 | Holland 071-125576 Fax : Holland 070-3280961 | -------------------------------------------------------------------------------- We apologize for the inconvenience.