Re: Change column name on DDL SQL table in iSeries Navigator

[CRPence <CRPbottle@xxxxxxxxx>]

  But from the perspective of implied triviality for the DB to be able 
to provide what the user wants....

  So do they not just correct the "source" of the dependents, the 
textual definitions stored in the catalogs, effectively using the 
following trivial scalar expression to rename the column just like would 
have happened for the column being renamed in the TABLE; thus there 
would be no reason to invalidate them, because those dependencies would 
have been corrected?:
   REPLACE(dependent_source, 'Old_Column_Name', 'New_Column_Name')

  Obviously it just *must be* that simple [for a database defined by 
text rather than by self-described database /objects/ like DB2 for i]! 
Even if they do so, then of course, it is not ever so simple :-)  It 
should be *obvious* for example, that the noted expression would /break/ 
any references to a column named as, for example, 
'The_Old_Column_NameX'; i.e. having become 'The_New_Column_NameX' even 
though that /rename/ effect was not also requested.

Regards, Chuck

On 21 May 2013 09:30, Mark Murphy/STAR BASE Consulting Inc. wrote:
> I suspect that they are invalidated because they would no longer
> match  their source if the changes were just cascaded.
>
> rob on 05/21/2013 10:45AM wrote:
>
> > The changes are not cascaded? The dependent objects are
> > "invalidated" instead?
> > Sounds like a hoot.
> >
> > Luis Rodriguez on 05/21/2013 10:41 AM wrote:
> >
> > > Of course, you must VERY careful when renaming columns, because
> > > any dependent objects (views, indexes, etc.) may be affected (in
> > > other DB2 versions, they are "invalidated").