Re: Reserved words

>From: gaffer@rec.com (Andrew Beattie) >Subject: Reserved words >Date: Thu, 11 Nov 1993 08:44:11 GMT >X-Informix-List-Id: <news.4859> > >Something that's been irritating me of late... > >ISQL will let me do this: > create table foo ( > absolute char(1), > accept char(1), > after char(1), > array char(1), > ascending char(1), > ascii char(1), > at char(1), > attribute char(1), > attributes char(1), > average char(1) > );> >But I think that you would agree that it is rather of a bad idea :-) >(This example is contrived, but it reflects a technique that I often use >for shooting myself in the foot). > >It would be rather nice if the back end refused to create tables or fields >with names that are reserved words in *any* Informix product (or at least >warned you that it was probably a bad idea). > >Any hope? Hi, Mr Beattie, No. Or rather, there are very few places where using a keyword as an identifier still causes trouble, and these are documented in places like the 4.10 supplement. Considerable effort was put into making keywords re-usable, largely to protect the innocents who have things like a column called REFERENCES -- an impeccable column name up until the 5.0 engines arrived on the scene. It also means that you should be able to design code (domain) tables with the structure: CREATE TABLE Domain (Code CHAR(n) NOT NULL PRIMARY KEY CONSTRAINT Pk_domain, Name CHAR(60) NOT NULL); where the keyword Name, used in I4GL, should not cause troubles. The keyword sensitivity was removed in 4.10 -- don't try it with 4.00. Incidentally, I think your table was too modest in its attempt to confuse. What's wrong with: CREATE TABLE Table ( Char DECIMAL(16,4), Null CHAR(10) NOT NULL, Serial DATE, Default CHAR(10) DEFAULT "default" NOT NULL, Decimal SERIAL(1000) NOT NULL PRIMARY KEY CONSTRAINT pk_table, Column CHAR(3) NOT NULL ); -- Yes, this works with a 5.01.UC2 database. But I tried adding a column -- called PRIMARY and that didn't work. I'd assume that a column called -- FOREIGN is also out of the question, and ditto UNIQUE and CHECK. Yours, Jonathan Leffler (johnl@informix.com) #include <disclaimer.h>