Re: fragmentation ? (fwd)

}From: inf_bb@hermes1.sps.mot.com (Bob Baskett) }Date: Fri, 9 Jun 1995 13:52:30 -0700 (MST) }X-Informix-List-Id: <list.6514> } }note on fragmentation....i've sent this to jonathan for help. And when I asked the senior R&D people about this, I got back the response: If the user intends to store numbers in the character field, then the specified fragmentation definition is perfectly legitimate, and therefore, we shouldn't be returning any errors. And, while I'll allow that it will mean that the fragment in dev_data_dbs will contain nothing because all numbers that can fit into a CHAR(1) field are less than 100, in general, the option to specifiy the ranges on a CHAR field as numbers could be useful if the data to be stored is to be numeric. So don't expect anything to change. Yours, Jonathan Leffler (johnl@informix.com) #include <disclaimer.h> }the following fragmentation sql statement creates the table but the load }fails due to the given error: } }CREATE TABLE "informix".fragbyexpnorem }( }c1 CHAR(1), }c2 SERIAL NOT NULL }) }FRAGMENT BY EXPRESSION }(c1 < 100 ) IN dev_root_dbs, }(c1 >= 100 ) IN dev_data_dbs }lock mode PAGE }; } }insert into fragbyexpnorem values ('a',0); }insert into fragbyexpnorem values ('b',0); } } 271: Could not insert new row into the table. } 883: Cannot evaluate the fragmentation expression. } }but if you change the fragmentation expression to : } }FRAGMENT BY EXPRESSION }(c1 < '100' ) IN dev_root_dbs, }(c1 >= '100' ) IN dev_data_dbs }lock mode PAGE }; } }it works. but shouldnt the engine check this ahead of time?