Re: Serial Update error

In article <beiehr$l0k$1@terabinaries.xmission.com>, "Murray Wood (IList)" <ifxmaillist@quanta.co.nz> writes > >Yes the same could happen with IDS under the same circumstances - probably >deadlock with 2 users updating 1 row in a table each but having to continue >their scan through the rest of the table. This would be considered a >database design issue, not an engine bug. The way the application is written precludes two users trying to update the same row - the program is locking a unique key in a second table to control access by the users, and that bit certainly works. I tested it to be sure! > >MW > >> -----Original Message----- >> From: owner-informix-list@iiug.org >> [mailto:owner-informix-list@iiug.org]On Behalf Of Five Cats >> Sent: Thursday, 10 July 2003 10:11 a.m. >> To: informix-list@iiug.org >> Subject: Serial Update error >> >> >> >> Found an interesting problem with a STANDARD ENGINE database. Note >> this, STANDARD ENDING *not* IDS. >> >> Table has a serial column. Program has 'UPDATE WHERE serial-column = >> program-variable'. Table does *not* have an index on the >> serial column, >> and starts producing locking errors presumably because it is scanning >> the whole table (yes I know that's bad) and coming across a >> row locked >> to another user. >> >> Now the table was missing an index which fixed the problem, but would >> this happen with IDS? Is it considered a bug in SE? >> >> Really this is curiosity on my part as due to the nature of >> the update >> there should be an index on the column, but curiosity can be a useful >> think. Of course especially with my sig I should note what it did to >> the cat as well.... -- Five Cats