Doug Chamberlin wrote:

> At 09:31 AM 11/9/2007, Adriano dos Santos Fernandes wrote:
>
>> 1) A common usage pattern of autonomous transaction will be to insert one
>> row in a table.
>>
>
> I would like to caution everyone about making such assumptions. Once a
> feature is added there will be every possible variation tried, often in
> ways we have not yet envisioned. Therefore, you CANNOT predict the "common
> usage pattern".
>
> If you want to add a feature in support of a specific usage pattern please
> try to do it in a way that limits or prevents variations and
> re-interpretations of the original intent. The perversity of user
> contortions has no bounds! Look at what people are always trying to do
> using UDF's!!
>
>

Hear this man! Features take on a life all their own. I've never been
able to predict what people will do with new features.

Actually, it is very instructive to see what people do. People almost
never ask for reasonable features. Rather, they ask for workarounds for
the absence of a feature. The classic example was about a trillion
years ago (circa 1977) when early Datatrieve users campaigned for a
"redefines" clause in the DDL a la Cobol. Turns out they really didn't
want redefines at all but a date data type, but nobody even considered
that. Redefines was how Cobol handled the problem, so that's what they
asked for. Nobody ever asked for blobs, either.

--
James Starkey, Senior Software Architect
MySQL Inc., Manchester, MA, USA, www.mysql.com
Office: 978 526-1376