Return-Path: pg_adm@postgres.berkeley.edu Received: by postgres.Berkeley.EDU (5.61/1.29) id AA13164; Mon, 27 Jul 92 06:58:26 -0700 Message-Id: <9207271358.AA13164@postgres.Berkeley.EDU> From: "Schenkelaars.V.F" Subject: Adding a new access method (long) To: postgres@postgres.berkeley.edu Sender: pg_adm@postgres.berkeley.edu To: postgres@postgres.berkeley.edu (Postgres mailing list) Date: Mon, 27 Jul 92 16:13:58 MET DST Cc: pjmg1@tnofel.fel.tno.nl (Peter van Oosterom), mao@postgres.berkeley.edu (Mike Olson) X-Mailer: ELM [version 2.3 PL11] Dear Postgres Group, I'm preparing to create a new access method for postgres. I`ve studied the R-tree code as provided with the postgres system and have come to teh following "recipe": The Reactive Tree is a modified R-tree so I can use the rtree functions as a template. I have to create the following functions for my access method the Reactive Tree: Reactgettuple (rtgettuple) Reactinsert (rtinsert) Reactdelete (rtdelete) Reactbeginscan (rtbegins) Reactrescan (rtrescan) Reactendscan (rtendscan) Reactmarkpos (rtmarkpos) Reactrestrpos (rtrestrpos) Reactbuild (rtbuild) These functions are registred in the pg_proc relation. My first question: Are these the only one I have to make or am I missing some functions? Next I append a tuple to pg_am. pg_am ( amname = char16 , <= "Reactive-tree" amowner = oid , <= is added automaticly amkind = char , <= "o" ? but what does this mean? amstrategies = int2 , <= Do I have to give the number of strategies here? btree has 5 but rtree does have 5 here too. I assumed the rtree has got 8 strategies, or am I wrong? amsupport = int2 , <= number of support functions: Union, Intersect and Size ? amgettuple = regproc , <= Reactgettuple aminsert = regproc , <= Reactdelete amdelete = regproc , <= Reactdelete amgetattr = regproc , <= - ? unknown to me amsetlock = regproc , <= - ? unknown to me amsettid = regproc , <= - ? unknown to me amfreetuple = regproc , <= - ? unknown to me ambeginscan = regproc , <= Reactbeginscan amrescan = regproc , <= Reactrescan amendscan = regproc , <= Reactendscan ammarkpos = regproc , <= Reactmarkpos amrestrpos = regproc , <= Reactrestrpos amopen = regproc , <= - ? unknown to me amclose = regproc , <= - ? unknown to me ambuild = regproc , <= Reactbuild amcreate = regproc , <= - ? unknown to me amdestroy = regproc <= - ? unknown to me ) Could yould answer the questionmarks in this table. I guess it just doesn`t aplly to the rtree access method. Next I have to create a opclass. But since the Reactive tree is similar to thre R-tree I just can use the R-tree opclasses. Or am I wrong? After this I should be able to define a Reactive Tree index: define index react-tree on test using Reactive-tree (bbox box_ops)` where test is a relation with attribute bbox of type box. My final question is: Is this the correct "recipe" for adding an access method. Or did I miss something. I don't want to waste a lot of time trying things that won't work. So I`ll be very pleased if you could answer my questions. 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.