Re: Attributes Names Standards

From: Craig Denson <denson_at_sdd.hp.com>
Date: 27 Jan 1994 13:31:47 -0800
Message-ID: <2i9bs3INNr7a_at_hpsdlss5.sdd.hp.com>


In article <jdennisCKAuy3.M0L_at_netcom.com> jdennis_at_netcom.com (John Dennis) writes:
>I trying to argue a point and need a little help. Some "micro-manager"
>wants to included a column type into the column name like : INT_AGE,
>indicating that AGE column has a type INT. This is crazy! What if
>someone decides that a smallint (sybase) would be better. Application
>code would have to be modified and recompiled, etc if this was to happen.
>So any solid reasons NOT to do this besides "its just no done"?

  i'm in the process of reading Codd's 1990 Relational Model, Version 2 and he seems to recommend doing exactly what your micro-manager is suggesting - with a twist. he suggests putting the domain type in the name, the domain type being a level up from the built in types. in your example the domain type might be years or age_in_years or something like that.

  this would isolate you from the lowlevel implementation - maybe. has anyone attempted this with sybase user-defined types or anything similar?

 seems like a pain to me - but then what do i know? craig Received on Thu Jan 27 1994 - 22:31:47 CET

Original text of this message