Re: Would be really nice if...

kirtan wrote:
> Using the same logic,
> 'select * from t' would also break any application if an alter table
> has been done to it.
> This would happen where the columns are being retrieved by position
and
> not by name.
> But ain't this feature still there?!

Such, in my mind, is not a 'feature' as much as a shorthand method for selecting all columns. And, yes, under the same scenario the application would still break with the same errors. However, most programmers know how 'select *' can kill an application for this very reason; adding this ridiculous appendage to 'select *' would make it, in my opinion, more widely used than it is already, and provide more avenues for application failure due to a misunderstanding of how it works and what may cause it to fail. I know programmers who 'know better' yet still code the ubiquitous 'select *' on occasion to speed the process along. Eventually ths application fails for the listed reason, and the programmer must recode the select correctly to prevent such errors in the future.

I also agree with Serge that most users really don't understand why things work and why some language constructs can create problems; they simply know it works for them at the moment, and happily go about their coding with nary a care. Bloating the language with this nonsense is, in my opinion, folly. And I've yet to see anyone coding a select with 97 columns who wasn't using either copy and paste or some query generating tool.

Sometimes 'innovative thinking' isn't so innovative, nor in the best interest of the language. This is one such case.

David Fitzjarrell Received on Thu Feb 24 2005 - 08:16:47 CST