Return-Path: postman Delivery-Date: Mon, 30 Aug 93 23:25:40 PDT Return-Path: postman Received: by postgres.Berkeley.EDU (5.61/1.29) id AA20698; Mon, 30 Aug 93 23:23:10 -0700 Resent-From: postman (POSTGRES mailing list) Resent-Message-Id: <9308310623.AA20698@postgres.Berkeley.EDU> Sender: owner-postman@postgres.Berkeley.EDU X-Return-Path: aoki@postgres.Berkeley.EDU Received: from faerie.CS.Berkeley.EDU by postgres.Berkeley.EDU (5.61/1.29) id AA20690; Mon, 30 Aug 93 23:23:03 -0700 Received: from localhost (localhost [127.0.0.1]) by faerie.CS.Berkeley.EDU (8.1C/8.1B) with SMTP id XAA12385; Mon, 30 Aug 1993 23:33:16 -0700 Message-Id: <199308310633.XAA12385@faerie.CS.Berkeley.EDU> From: aoki@postgres.Berkeley.EDU (Paul M. Aoki) To: Nabiel Al-Madani Cc: postgres@postgres.Berkeley.EDU Subject: Re: Postquel queries In-Reply-To: Your message of Mon, 30 Aug 93 22:36:21 EDT <9308310236.AA13655@zeus.csee.usf.edu> Date: Mon, 30 Aug 93 23:33:15 -0700 X-Sender: aoki@postgres.Berkeley.EDU X-Mts: smtp Resent-To: postgres-dist Resent-Date: Mon, 30 Aug 93 23:23:09 PDT general notice. i can't promise that all bugs reported here will be fixed (and actually, bugs should be reported to bug-postgres), but i can promise that from now on i (and only i) will throw away any bug reports that do not contain (at an absolute minimum) the version of postgres and the operating system/architecture. someone else here may very well look at it, but i won't. Nabiel Al-Madani writes: > I have two questions concerning Postquel queries. The first is, are there any > floating point binary operators defined in postgres? yes. > emp is defined as: emp(ss_no = int4, salary = float4), and postgres responded > by the following message: > Query sent to backend is "retrieve (emp.all) where emp.salary > 35000.00" > NOTICE:Aug 27 04:09:31:there is no operator > for types float4 and float8 > NOTICE:Aug 27 04:09:31:You will either have to retype this query using an > NOTICE:Aug 27 04:09:31:explicit cast, or you will have to define the operator > WARN:Aug 27 04:09:31:> for float4 and float8 using DEFINE OPERATOR > if there is none, then how can I explicitly cast the query. from the manual: Constants of POSTGRES User-Defined Types A constant of an arbitrary type can be entered using the notation: "string"::type-name In this case the value inside the string is passed to the input conver- sion routine for the type called type-name. The result is a constant of the indicated type. hence, retrieve (emp.all) where emp.salary > "35000.00"::float4 should work.. (if it doesn't there's something REALLY wrong.. ^_^;) avi pfeffer recently cleaned up the typecasting/coercion code so that smarter things are done more often and ambiguous cases are handled more consistently. (it was pretty ad hoc before..) [query with a special kind of 'or' clause] > Postgres responded by the following message: > Error: No response from the backend, exiting... yep, core-dump-o-rama. gotta hate those existential/free variables. i *think* the following context diff shows the correct fix, relative to postgres 4.1. (save the original in case you have to back it out :-) however, because of the funky duplicate semantics of postquel, you had better change your query to: retrieve unique (student.all) where student.Id = TA.Id or student.Id = RA.Id\g unless you want to see tons of duplicates. *** 1.17 1992/11/23 22:38:29 --- src/backend/planner/path/joinrels.c 1993/08/31 06:00:38 *************** *** 103,113 **** get_rel(CAR(other_rels)), joininfo), LispNil); ! else temp_node = lispCons((LispValue)init_join_rel(outer_rel, get_rel(other_rels), joininfo), LispNil); t_list = nconc(t_list,temp_node); } --- 103,115 ---- get_rel(CAR(other_rels)), joininfo), LispNil); ! else if (BushyPlanFlag) temp_node = lispCons((LispValue)init_join_rel(outer_rel, get_rel(other_rels), joininfo), LispNil); + else + temp_node = LispNil; t_list = nconc(t_list,temp_node); } -- Paul M. Aoki | CS Div., Dept. of EECS, UCB | aoki@postgres.Berkeley.EDU | Berkeley, CA 94720 | ...!uunet!ucbvax!aoki