Re: OO versus RDB

On 5 Jul 2006 08:47:55 -0700, "topmind" <topmind_at_technologist.com> wrote:

>J M Davitt wrote:

[snip]

>> There should be no risk associated with "name refactoring." One

     "should"? Why? Is it impossible or morally wrong?

>> should not have to "try to hunt down all references." If you have
>> to hunt and risk missing something, you're not managing your code.
>
>First of all, it may not be "my code". Second, there may be many
>languages and tools that don't readily allow easy substitution of
>column names. For example, many point-and-click tools don't expose
>their innards to such searching. Any kind of meta programming also may
>hide such from search. Whether these tools are good or bad is another
>issue; such tools exist.

     Third, it may be generated code. The in-house client billing system that I have developed and maintain has generated code in the source. (Yes, there are warnings not to change the generated code.) To do it right involved also editing another level of source and regenerating the source.

     Fourth, the column name's identifier may be used elsewhere as something else. Are you sure that, say, "charge" here is the column charge?

[snip]

Sincerely,

Gene Wirchenko Received on Wed Jul 05 2006 - 21:52:22 CEST